Radware, Ltd. v. F5 Networks, Inc.

*979ORDER ON: (1) RADWARE’S MOTION TO STRIKE; (2) F5’S MOTION FOR JUDGMENT ON THE PLEADINGS; (3) F5’S MOTION TO AMEND INVALIDITY CONTENTIONS; (4) F5’S MOTION FOR SUMMARY JUDGMENT OF INVALIDITY; (5) RADWARE’S MOTION FOR PARTIAL SUMMARY JUDGMENT ON INVALIDITY; (6) RADWARE’S MOTION FOR SANCTIONS; (7) RADWARE’S MOTION FOR SUMMARY JUDGMENT OF INFRINGEMENT; (8) F5’S MOTION FOR SUMMARY JUDGMENT OF NON-INFRINGEMENT; (9) F5’S MOTION FOR SUMMARY JUDGMENT ON DAMAGES

Ronald M. Whyte, United States District Judge

Defendant F5 Networks, Inc. (“F5”) and plaintiffs Radware, Inc. and Radware Ltd. (collectively “Radware”) bring the following motions: (1) Radware’s Motion to Strike, Dkt. No. 201; (2) F5’s Motion for Judgment on the Pleadings, Dkt. No. 180; (3) F5’s Motion to Amend Invalidity Contentions, Dkt. No. 212; (4) F5’s Motion for Summary Judgment of Invalidity, Dkt. No. 183; (5) Radware’s Motion for Partial Summary Judgment Motion Against F5’s Affirmative Defense of Invalidity, Dkt. No. 189; (6) Radware’s Motion for Sanctions, Dkt. No. 207; (7) Radware’s Motion for Summary Judgment of Infringement, Dkt. No. 182; (8) F5’s Motion for Summary Judgment of Non-Infringement, Dkt. No. 190; and (9) F5’s Motion for Summary Judgment on Damages Issues, Dkt. No. 187. The court has reviewed the papers-filed and heard the argument of counsel. The court rules on the motions as set forth below.

I. BACKGROUND

A. Asserted Patents

Radware brings this patent infringement action against its competitor F5, alleging infringement of claims 1-7, 9-19, and 21-32 of U.S. Patent No. 8,266,319 (“’319 Patent”) and claims 1-4, 6-12, 14, and 15 of U.S. Patent No. 8,484,374 (“’374 Patent”) (collectively “Asserted Patents”).1 Both Asserted Patents are entitled “Load Balancing” and relate to “computer networks in general, and in particular to load balancing client requests among redundant network servers in different geographical locations.” ’319 col.l 11.13-16; ’374 col.l II.17-20. The ’374 Patent is a continuation of the ’319 Patent. The ’374 Patent shares the same specification as the ’319 Patent, other than the “Summary” section.

The technology at issue relates to link load balancing in a multi-homed environment. A “multi-homed” network is a network with multiple connections to the Internet. ’319 col.15 11.34-37. “Link load balancing” is a process for allocating network communications across these connections.

The Asserted Patents relate to techniques and systems for selecting a specific route from the multi-homed network to the Internet and from the Internet into the multi-homed network. The claimed inventions describe both “outbound” and “inbound” link load balancing. Outbound link load balancing deals with requests sent from a host that are destined for an external network via the Internet. Inbound link loading involves inbound requests for services received by the host via the Internet. Claims 24-28 of the ’319 Patent are directed to outbound link load balancing. Claims 1-23 and 29-32 of the ’319 Patent and all *980claims of the ’374 Patent are generally-directed to inbound link load balancing.

The Asserted Patents claim link load balancing as both a method and system. Representative Claim 26 of the ’319 Patent describes a method for outbound link load balancing: ,

26. A method for routing data via a network from a first node to a second node, said network having a plurality of available routes from said first node to said second node and the plurality of routes are assigned with respective IP addresses, said method comprising the steps of:

selecting one of said routes for sending data between the first node and the second node on the basis of costing information of said- respective routes; 1 ”

receiving a packet having a source IP address;, and

translating the source^ IP address to an IP address corresponding to the selected route of the plurality of routes...

Representative Claim 13 of the ’319 patent describes a method for inbound link load

balancing:

13. A method for managing a computer network having a device connected to the Internet through a plurality of routes, wherein the plurality of routes are assigned with respective IP addresses, comprising:

receiving a DNS2 resolution query from a remote computer for a domain name within the computer network; selecting one of a plurality of routes connecting said device to the Internet in accordance with one or, more criteria of the plurality of routes;

responding to the DNS resolution query with an IP address associated with the selected route, said IP address is used for resolution of said domain name, receiving a packet having a destination IP address corresponding to one of the plurality of routes; and

translating the destination IP address to an IP address within the computer network.

B. Accused Products

Radware accuses F5’s “BIG-IP Application Delivery Controller” of infringement. Specific accused models are listed at Dkt. No. 182 at 10 n.4. F5’s BIG-IP device is [redacted] Dkt. No. 179-4 at 10 (citing Stamm Rep.). The infringement issues focus on three modules within the TMOS: the Link Controller, Local Traffic Manager (“LTM”), and Global Traffic Manager (“GTM”). Id. at 10-11. All three [redacted] must be activated through the purchase of á license from F5. Id. at 11. “LTM’s primary functionality is local server load balancing, not ISP load balancing. In particular, an LTM can sit between a local network arid the internet, and control the routing of incoming messages to different servers.” Dkt. No. 184-4 (Brewer Deck) ¶ 8. “GTM’s primary functionalities are to provide DNS-related services and global server load balancing (‘GSLB’). DNS services relate to ■ resporiding to client ■ requests for IP addresses associated with a domain name (e.g., ‘Amazon.com’).” Id. ¶ 9.

F5 acknowledges that the Link Controller’s “primary purpose is to provide, in part, ISP link load balancing functionality.” Dkt. No. 190 at 3. However, in December 2014 F5 implemented a “hotfix” that made certain changes to its source code for the LTM, GTM and Link Controller that, according to F5, [redacted] Id. at 4; see also Dkt. No. 188-6 (Thornewell Deck) *981¶ 11. Since removing the Link Controller load balancing functionality, F5 has [redacted] Brewer Decl. ¶ 16.

C. Procedural History

The court previously held a joint Mark-man hearing in this ease and in the related case Radware v. A10, 13-cv-2021, on April 8, 2014 and issued its Claim -Construction Order on April 18, 2014. Dkt. No. 122. The court also considered motions for summary judgment related to .claim construction issues in May 2015, and issued an order on the motions on June 11, 2014. Dkt. No. 145. The parties in the A10 case filed a Joint Notice of Settlement on August 29, 2014.13-cv-2021, Dkt. No. 252.

Having held a healing on July 24, 2015, the court addresses the parties’ various motions: (1) Radware’s Motion to Strike, Dkt. No. 201, F5’s Motion for Judgment on the Pleadings, Dkt. No. 180, and F5’s Motion to Amend Invalidity Contentions, Dkt. No. 212; (2) F5’s Motion for Summary Judgment of Invalidity, Dkt. No. 183, and Radware’s Motion for Partial Summary Judgment Motion Against F5’s Affirmative. Defense of Invalidity, Dkt. No. 189; (3) Radware’s Motion for Sanctions, Dkt. No. 207; (4) Radware’s Motion for Summary Judgment of Infringement, Dkt. No. 182, and F5’s Motion for Summary Judgment of Non-Infringerrtént, Dkt. No. 190; and (5) Motion for Summary Judgment on Damages Issues, Dkt. No. 187.

II. RADWARE’S MOTION TO STRIKE PORTIONS OF F5’S MOTION FOR MOTION FOR JUDGMENT ON THE PLEADINGS, SUMMARY JUDGMENT OF INVALIDITY, AND DECLARATION OF DR. ALEXANDER IN SUPPORT THEREOF

Radware moves to strike: (1) F5’s motion for judgment on the pleadings; (2) those portions of F5’S motion for summary judgment of invalidity that rely on Global Server Load Balancing (“GSLB”); ■ (3) those portions of F5’s motion for summary judgment of invalidity that rely on Border Gateway Protocol (“BGP”); and (4) the May 28, 2015 declaration of Dr. Peter Alexander submitted in support of F5’s motion for summary judgment of invalidity. Dkt. No. 201.

A. F5’s Motion for Judgment on the Pleadings and its Section 101 Invalidity Defense

' Radware moves to strike F5’s motion for judgment on the pleadings on the basis that F5 failed to disclose 35 U.S.C. § 101 subject matter ineligibility as a basis for finding the Asserted Patents invalid in its invalidity contentions. Id. at 1.

F5 served its preliminary invalidity contentions on September 30, 2013, Dkt. No. 189-9, and, after obtaining the court’s leave, see Dkt. No. 157, served amended invalidity contentions on May 16, 2014, Dkt. No. 189-10. The preliminary, invalidity contentions state only that: (1) several asserted claims are invalid under 35 U.S.C. § 101 for mixing statutory claim classes, Dkt. No. 189-9 at 59 n.l, and that because discovery had then only recently commenced, it was at that time “premature for F5 to determine whether there are issues related to... improper inventorship [and] derivation under 35 U.S.C. § 101/102(f).” Id. at 59. F5’s amendments did not add any invalidity contentions related to Section 101. Only after Radware filed the instant motion to strike F5’s Section 101 invalidity defense did F5 move to amend its invalidity contentions to include the defense of lack of subject matter eligibility. Dkt. No. 212.

Under Patent L.R. 3-3(d), a party’s invalidity contentions must disclose “[a]ny grounds of invalidity based on 35 U.S.C. § 101... of any of the asserted claims.” “Any invalidity theories not disclosed pursuant to Local Rule 3-3 are barred, accord*982ingly, from presentation at trial (whether through expert opinion.testimony or otherwise),” MediaTek Inc. v. Freescale Semiconductor, Inc., Case No. 11-5341, 2014 WL 690161, at *1 (N.D.Cal. Feb. 21, 2014) (citation omitted). This District’s patent local rules are designed “to balance the right to develop new information in discovery with the need for certainty as to the legal theories.” O2 Micro Int’l Ltd. v. Monolithic Power Sys., Inc., 467 F.3d 1355, 1366 (Fed.Cir.2006). The rules-accomplish this by requiring parties to “provide early notice of their,.. invalidity contentions, and to proceed with diligence in amending those contentions when new information comes to light in the course of discovery.” Id. at 1365-66. As Judge Gre-wal stated in a recent decision denying a motion for judgment on the pleadings for failure to disclose Section 101 invalidity as required by Patent L.R. 3-3, “[t]his district’s patent rules only work if there are consequences for failing to comply.” Good Technology Corp. v. MobileIron, Inc., Case No. 12-5826, Dkt. No. 298 at *3, 2015 WL 3866019 (N.D.Cal. May 4, 2015).

F5 does not dispute that it failed to disclose Section 101 as an invalidity theory as required under the Patent Local Rules. Rather, F5 argues that the court should deny Radware’s motion to strike F5’s Section 101 defense because: (1) the defense is based on new legal authority developed within the last year, subsequent to the Supreme Court’s decision in Alice Corp. v. CLS Bank Int’l, — U.S. —, 134 S.Ct. 2347, 189 L.Ed.2d 296 (2014); and (2) the defense is not based on claim construction or'fact issues subject to discovery, so Rad-ware will not suffer any prejudice should the . court permit F5’s Section 101 defense. Dkt. No, 213 at 2. Neither argument is persuasive.

As to the first, F5 waited too long following Alice to seek amendment - of its invalidity contentions to assert a Section 101 defense, and an intervening district court application of Alice does not. constitute sufficient change in the law to permit amendment so late in the case. Intervening change in the law can constitute good cause to amend invalidity contentions. F5 cites a decision from the Central District of California permitting the amendment of claims in light-of Alice. Dkt. No. 213 at 3. However, the defendant in that case served its amended invalidity contentions on August 15, 2014, less than two months following Alice. See Mortgage Grader, Inc. v. Costco Wholesale Corp., Case No. 13-0043, Dkt. No. 97 at 2; 2014 WL 10763261 (C.D.Cal. Oct. 27, 2014). Here, F5 waited far longer — nearly a year — before filing its motion for judgment- on the pleadings asserting invalidity under Section 101. In addition, the issue of what constitutes patentable-subject matter'has been hotly debated since the Federal Circuit’s 2013 en banc decision in CLS Bank Intern. v. Alice Corp. Pty. Ltd., 717 F.3d 1269 (Fed.Cir.2013), aff'd, — U.S. —, 134 S.Ct. 2347, 189 L.Ed.2d 296 (2014).

F5 also bases its motion on Intellectual Ventures v. JP Morgan Chase & Co., Case No. 13-3777, 2015 WL 1941331 (S.D.N.Y. April 28, 2015). See Dkt. No. 180. According to F5, Intellectual Ventures, as “the first post-Aiiee decision addressing patent-ability of networking methods and technologies,” constitutes “intervening law” which supports late amendment to assert a Section 101 defense. Dkt. No. 213 at 3. However, Intellectual Ventures is not controlling authority: Alice marked the shift in Section 101 invalidity analysis; Intellectual Ventures merely applied Alice to the field of networking méthods and technologies. Accordingly, Intellectual Ventures does not constitute the kind of “intervening law” that would support amendment at this late hour. As Radware notes, the defendant in Good Technology submitted the Intellectual Ventures decision in support *983of its motion for judgment on the pleadings asserting a previously undisclosed Section 101. defense. See Good Technology Corp., Case No. 12-5826, Dkt. No. 281-1. Despite the Intellectual Ventures decision being' even more recent at that point than it is now, Judge Grewal nonetheless refused to excuse noncompliance with the Patent Local Rules and allow a previously undisclosed Section 101 theory. See id., Dkt. No. 298;

As to F5’s second ’ argument, the court need not reach the question of prejudice to Radware because F5 has not shown good cause to permit amendment of it's invalidity contentions. Patent L.R. 3-6'pro-vides for amendment of invalidity contentions “only by order of the Court upon a timely showing of good cause.”' “[G]ood cause requires a showing of diligence,” and it is the movant’s burden to-establish diligence. O2 Micro, 467 F.3d at 1366. As discussed above, F5- has not shown diligence, so the court need not consider the prejudice to Radware that would flow from F5’s untimely assertion' of á Section 101 invalidity theory.

Radware’s motion to strike F5’s motion for judgment on the pleadings is GRANTED, and' F5’s motion for judgment on the pleadings is DENIED. For the same reasons, F5’s motion to amend its invalidity contentions to add Section 101 subject matter ineligibility as a defense is DENIED.

B. Global Server Load Balancing and Cisco DistributedDirector

Radware also moves to strike those portions of F5’s motion for summary judgment of invalidity that rely on Global Server Load Balancing (“GSLB”) because F5 failed to disclose GSLB as prior art in its invalidity contentions as required by Patent Local Rule 3-3(c). Dkt. No.' 201 at 3.

F5 does not dispute that the term GSLB does not appear in- its invalidity contentions but states that its summary judgment motion of invalidity is based on a specific embodiment of GSLB, Cisco’s Dis-tributedDirector (“Ciscó DD”), not the general concept of GSLB itself. Dkt. No. 213 at 4. According to'F5, all aspects of GSLB relied upon in its summary judgment motion are disclosed by Cisco DD, and Cisco DD forms the entire basis for its analysis regarding invalidity based on GSLB. Id. at 5. F5 asserts that it both identified and charted Cisco DD as invalidating prior art in both its preliminary invalidity contentions and in Dr. Alexander’s invalidity report.'/^, at 6.

Radware, while acknowledging that Cisco DD was previously disclosed, nevertheless argues that “any" argument that GSLB, other than [Cisco DD], constitutes invalidating prior art should be stricken,” and that F5 “should be precluded from referring to GSLB functionality as invalidating prior art, whether or not it is a functionality performed by [Cisco DD].” Dkt. No. 224 at 5. However, Radware leaves unrebutted F5’s contention that Cisco DD is a specific embodiment of GSLB and cites no authority for precluding mention of GSLB as implemented in Cisco DD. While some of the language in F5’s motion does imply that GSLB systems other than Cisco DD also constitute anticipating prior art, see; e.g., Dkt. No. 183 at 14 (“The asserted claims are invalid based on GSLB systems, such as the Cisco DistributedDi-rector. ...”), Radware identifies no portion of F5’s motion for summary judgment of invalidity that relies on an aspect of GSLB not disclosed by Cisco DD.

Accordingly, because F5’s motion for summary judgment is based on Cisco DD, rather than GSLB generally, and Cisco DD was previously disclosed as allegedly anticipating prior art, Radware’s motion to *984strike those portions of F5’s motion based on GSLB is DENIED.3

C. Border Gateway Protocol

Radware moves to strike those portions of F5’s motion for summary judgment of invalidity that rely on BGP because F5 failed to disclose BGP as prior art in its invalidity contentions. Dkt. No. 201 at 3.

Like GSLB, BGP is a general protocol, rather than a specific implementation or embodiment. However, whereas F5’s motion for summary judgment makes clear that Cisco DD is a specific implementation of GSLB and that Cisco DD itself is the prior art reference upon which its invalidity. theory is based, F5 does not dispute that it failed to disclose BGP as a prior art reference and has not identified a specific BGP reference. Instead, F5 argues that BGP was discussed in both parties’ expert reports, and that Radware cannot' claim that it will suffer any undue prejudice from its consideration as prior art. Dkt. No.' 213 at 7. F5 states that its’summary judgment motion regarding BGP is “based entirely on Radware’s expert’s own admissions from his April 17, 2015 deposition,” and that “[pjrior to conducting the deposition, F5 could not have anticipated that Dr. Rubin would make these admissions, and thus could not have presented them in its contentions.” Id.

F5 does not explain precisely how the testimony of Dr. Izhak Rubin, Radware’s technical expert, somehow enabled its BGP invalidity argument.4 Based on F5’s. argument, it appears that F5 either: (1) knew BGP was anticipating prior art before Dr. Rubin’s deposition but chose not to disclose it as á prior art in its invalidity contentions, perhaps because .before Dr. Rubin’s alleged ádmissions F5 felt it was an argument not worth making; or (2) F5 was aware of BGP but unaware that it was an, anticipating prior art reference until Dr. Rubin’s deposition. Whatever the case, F5 has not shown good cause to excuse its failure to comply with Patent Local Rule 3-3(c). F5 cites no authority for-the proposition that an opposing expert’s admissions can constitute good cause to allow the assertion of previously undisclosed prior art, and even if it could, F5 has not explained how Dr. Rubin’s alleged admission somehow rendered BGP. anticipating prior art that F5 could not have discovered in an earlier diligent search. This is particularly *985true in light of the fact that F5’s expert discussed BGP in his expert report.

As with F5’s Section 101 invalidity argument, because F5 has not shown good cause amend its invalidity contentions, the court need not reach the question of prejudice.

F5’s invalidity contentions do not disclose BGP as anticipating prior art as required under Patent Local Rule 3-3, and F5 has not shown good cause to amend its invalidity contentions. Radware’s motion to strike those portions of F5’s motion for summary judgment of invalidity based on BGP is therefore GRANTED.

D. May 28, 2015 Supplemental Declaration of Dr. Alexander

Finally, Radware moves to strike the supplemental declaration of F5’s expert, Dr. Alexander. Dkt. No. 201 at 12. Radware argues that in his supplemental declaration Dr. Alexander provides additional opinions regarding the Cisco DD system that were not provided in his expert report as required by Rule 26. Dr. Alexander provided his expert report on invalidity on January 20, 2015. In it, he identified and explained the Cisco DD testbed system he had created as a demonstration that Cisco DD anticipates the Asserted Patents. On March 2, 2015 Dr. Rubin provided his rebuttal expert report on invalidity, pointing out what are, in his opinion, flaws with the test-bed system. Specifically, Dr. Rubin opines that: (1) the test bed was not connected to the Internet, as required by the claims; (2) certain components of the test bed (a networking switch and configuration files) were not in existence as of the priority date of the patents; and (3) the test bed system, as configured by Dr. Alexander, still could not perform inbound or outbound link load balancing as required by the asserted claims. Dkt. No. 189-29 ¶¶ 96-109. Rad-ware contends that because Dr. Alexander created the test-bed, he had all of the evidence at his disposal when he wrote his initial expert report. Dkt. No. 201 at 13. According to Radware, to allow Dr. Alexander’s supplemental declaration would prejudice Radware by permitting F5 an improper rebuttal to Dr. Rubin’s rebuttal.

F5 contends that the supplemental declaration contains no new opinions, but rather simply addresses the. purported flaws with the test-bed system, which are simply factual issues raised by Radware. Dkt. No. 213 at 8. F5’s position is' that an expert cannot be reasonably expected to anticipate every counterargument that could possibly be made and preemptively discuss them in his report. Id, Moreover, F5 argues that Radware has had substantial opportunity to conduct discovery into these alleged flaws, including deposing Dr. Alexander about the substance of the

declaration.

The court agrees, with F5. Radware identifies no prejudice other than the general assertion that “it would prolong Rad-ware’s day in Court and the cessation of F5’s infringing conduct.” Dkt. No. 201 at 14. Moreover, the court agrees that an expert cannot reasonably be expected to anticipate every criticism and argument that an opposing expert will level, and preemptively respond to each and every one. Finally, the court notes that Radware cites no authority in. support of its argument, beyond citing Rule 26 itself. Accordingly, the court DENIES Radware’s motion to strike the supplemental declaration. However, F5 cannot offer new opinions of Dr. Alexander that go beyond merely responding to Dr. Rubin’s criticisms of the test bed systems.

*986III. THE PARTIES’ MOTIONS FOR SUMMARY JUDGMENT OF INVALIDITY AND TO DISMISS F5’S INVALDITY AFFIRMATIVE DEFENSES

Each party moves for summary judgment regarding invalidity, F5 moves for summary judgment that the asserted claims are invalid as anticipated by prior art GSLB and BGP, or, in the alternative, that the asserted claims are obvious in light of those references. Dkt. No. 183. Radware moves for partial summary judgment against F5’s affirmative defense of invalidity on the grounds that F5 has failed to demonstrate that the asserted references constitute invalidating prior art. Dkt. No. 189.

A. Legal Standard

Summary judgment is proper where the pleadings, discovery, and affidavits demonstrate that there is “no genuine issue as to any material fact and that the moving party is entitled to judgment as a matter of law.” Fed. R. Civ. P. 56(c); see also Celotex Corp. v. Catrett, 477 U.S. 317, 322, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986). At the summary judgment stage, the Court “does not assess credibility or weigh the evidence, but simply determines whether there is a genuine factual issue for trial.” House v. Bell, 547 U.S. 518, 559-60, 126 S.Ct. 2064, 165 L.Ed.2d 1 (2006). Material facts are those which may affect the outcome of the case. Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 248, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986). A dispute as to a material fact is genuine if there is sufficient evidence for a reasonable jury to return a verdict for the nonmoving party. Id.

Patents are presumed to be valid. 35 U.S.C. § 282(a). However, an invention is not patentable if “(a) the invention was known or used by others in this country, or patented or described in a printed publication in this or a foreign country, before the invention thereof by the applicant for patent, or (b) the invention was patented or described in a printed publication in this or a foreign country or in public use or on sale in this country, more than one year prior to the date of the application for patent in the United States” or if “(e) the invention was described in (1) an application for patent, published under section 122(b), by another filed in the United States before the invention by the applicant for patent or (2) a patent granted on an application for patent by another filed in the United States before the invention by the applicant for patent....” 35 U.S.C. § 102(a), (b), (e) (2006). A party challenging the validity of a patent bears the burden of proving invalidity by clear and convincing evidence. Microsoft Corp. v. i4i Ltd. P’ship, 564 U.S. 91, 131 S.Ct. 2238, 2242, 180 L.Ed.2d 131 (2011).

B. Prior Art References

Radware moves for partial summary judgment that: (1) F5 abandoned a number of prior art references previously cited in its invalidity contentions; (2) the “test-bed” systems created by F5 are not prior art; (3) the Maki-Kullas patent is not prior art; and (4) the Cisco DistributedDirector White Paper (“DD white paper”) is not prior art. See Dkt. No. 189. The court first addresses the allegedly abandoned prior art references and test-bed systems before turning to the Maki-Kullas patent and the DD white paper.

I. Abandoned Prior Art References

Radware first moves for partial summary judgment on twenty-six alleged anticipating prior art references and nine alleged obviousness combinations. Id. at II. Radware argues that despite identifying these prior art references and combinations in its invalidity contentions, F5’s expert failed to offer any opinions based *987on these references. Id. The result, according to Radware, is that F5 has no evidence in support of its claim that these references and combinations invalidate the Asserted Patents, and therefore Radware is entitled to partial summary judgment that these unsupported references do not invalidate the Asserted Patents, Id. at 13.

F5 does not dispute that Dr. Alexander did not rely or offer an opinion on these references but argues that “a district court should not exclude prior art evidence simply because such evidence is not in an expert’s report.” Dkt. No. 196-4 at 18 (citing Meyer Intellectual Props. Ltd. v. Bodum, Inc., 690 F.3d 1354 (Fed.Cir.2012)). F5 agrees that it may not raise such prior art at trial through expert testimony but argues that regardless of whether F5 intends to use the twenty-six prior art references and nine combinations, there is no basis for granting summary judgment in Radware’s favor.

F5 has not necessarily abandoned the asserted prior art references by failing to introduce any expert opinions based on these references. See In re Brimonidine Patent Litig., 643 F.3d 1366, 1376 (Fed.Cir.2011) (“There is no invariable requirement that a prior art reference be accompanied by expert testimony.”). Accordingly, Radware’s motion for summary judgment that F5 abandoned the prior art references on which it did not offer an expert opinion is DENIED. However, Radware may move to exclude these references if F5 later seeks to introduce them through lay witness testimony, should the particular references F5 seeks to introduce contain matter that is complex and requires expert testimony to explain. See, e.g., Alexsam, Inc. v. IDT Corp., 715 F.3d 1336, 1347-48 (Fed.Cir.2013). In addition, Radware may move to limit the number of prior art references asserted as anticipating or as part of an obviousness combination.

2. Cisco and BIG-IP Test-Bed Systems

Radware also moves for partial summary judgment that F5’s Cisco DD and BIG-IP test-bed systems do not anticipate or render obvious the asserted claims as a matter of law. Dkt. No. 189 at 14. Radware advances two arguments: (1) the test-bed systems did not exist as configured before the priority date of the patents because they incorporate components and configuration files that did not exist as of the priority date of the patents; and (2) the test-bed systems were not known or used to practice link load balancing before the priority date of the patents. Id. However, F5 appears to agree with Radware that the test-bed systems, rather than the underlying 3-DNS and DistributedDireetor products, are not prior art. See Dkt. No. 196-4 at 16 (“F5 has never asserted that all of the components of the Test-Bed System were used together in the prior art. Instead, the Test-Bed System simply evidences that the prior art 3-DNS and Dis-tributedDirector systems could perform ISP load balancing.”). Accordingly, because it is undisputed that the test-bed systems are themselves not prior art, Rad-ware’s motion for summary judgment that the test-bed systems neither anticipate nor render obvious the asserted claims is GRANTED.5

3. The Maki-Kullas Patent

Radware moves for summary judgment that the Maki-Kullas patent is not *988prior art. See Dkt. No. 189. The Maki-Kullas patent, filed on October 5, 1999, pre-dates the December 20, 1999 effective filing date of the ’819 (and ’374) .patents.6 However, the person “who first conceives, and, in a mental sense, first invents.. .may date his patentable invention back to the time of its conception,, if he connects the conception with its reduction to practice, by reasonable diligence on his part, so that they are. substantially one continuous act.” Mahurkar v. C.R. Bard, Inc., 79 F.3d 1572, 1577. (Fed.Cir.1996) (internal quotation marks and citation omitted). Accordingly, to overcome the Maki-Kullas prior art reference, Radware must demonstrate that it conceived of the inventions disclosed in the ’319 and ’374 patents prior to the Maki-Kullas priority date, October 5, 1999, and diligently reduced the inventions to practice after conception.7 Radware asserts that the invention disclosed .in thé ’319 and ’374 patents was conceived by April i, 1999, and reduced to practice at least as eárly as October 4, 1999. Dkt. No. 186:4 at 17-19.

Although F5 has the burden of establishing invalidity, Radware has the burden of. proving that it conceived of the alleged inventions before the filing date of the patent application and diligently reduced the inventions to practice. Taurus IP, LLC v. DaimlerChrysler Corp., 726 F.3d 1306, 1322 (Fed.Cir.2013), (“After an accused infringer has put forth a prima facie case of invalidity, the burden of production shifts, to the patent owner to produce sufficient rebuttal evidence to prove entitlement to an earlier invention date.”). To do so, Radware must corroborate its alleged conception and reduction to practice dates. Spectralytics, Inc. v. Cordis Corp., 576 F.Supp.2d 1030, 1045-46 (D.Minn.2008), aff'd, 649 F.3d 1336 (Fed.Cir.2011) (“The patentee must do more than offer some evidence (such as an affidavit) supporting his pre-patent application invention date; the patentee must offer corroboration of that invention date.”) (emphasis in original). “The issue of the conception date of an invention is a legal conclusion" based on underlying factual findings.” Taurus IP, 726 F.3d at 1322.

a. Conception

Radwaré" asserts that the invention disclosed in the ’319 and ’374 patents was conceived by April 1, 1999, see Dkt. No. 186-4 at 17-19, and offers the following evidence in support of this date:

• [redacted]

[redacted]

[redacted]

[redacted]

As an initial matter, Dr. Rubin’s expert testimony is based solely on his review of “the inventor testimony and various documents cited and referenced in the depositions.” Id.8 Radware does not specifically identify the documents reviewed by Dr. Rubin, but the documents appear to include only the above inventor testimony and the [redacted]. As a result, Dr. *989Rubin’s expert testimony adds nothing beyond the inventor testimony in this case, and does not constitute evidence of conception. See Invitrogen Corp. v. Clontech Labs., Inc., 429 F.3d 1052, 1068 (Fed.Cir.2005) (rejecting expert testimony in support of conception that simply pointed to the inventor’s notebook and concluded that various entries demonstrated conception without providing any further explanation).9

What this leaves is the testimony of Zisapel and Peles,* co-inventors- of the Asserted Patents-, and the email from Zisapel to Fuks, who is also a co-inventor of the asserted patent. The testimony of Zisapel and Peles cannot alone show conception, and must be corroborated. See, e.g., Rosco, Inc. v. Mirror Lite Co., 304 F.3d 1373, 1382 (Fed.Cir.2002) (“It is well established that a party claiming his own prior inventorship must proffer evidence corroborating his testimony.”). When evaluating corroboration, courts apply a “rule of reason” that examines “all pertinent evidence so that a sound determination of the credibility of the inventor’s story may be reached.” Coleman v. Dines, 754 F.2d 353, 360 (Fed.Cir.1985). Although under a rule of reason analysis all the pertinent evidence is considered, it “does not dispense with the requirement for somé evidence of independent corroboration.” Coleman, 754 F.2d at 360. Corroboration' of inventor testimony requires evidence beyond an inventor’s own statements and documents. See Hahn v. Wong, 892 F.2d 1028, 1032 (Fed.Cir.1989) (“the inventor. . .must provide -independent corroborating evidence in addition to his own statements and documents.”); Golden Bridge Tech. Inc. v. Apple, Inc., Case No. 12-4882, 2014 WL 1928977, at *4 (N.D.Cal. May 14, 2014) (“[E]vidence of conception and reduction to practice must be ‘independently corroborated’ beyond the inventor’s own statements and documents, and must be ‘evidence that would-be available to a jury’ to support its factual positions.”) (internal quotation marks and citation omitted).

Even viewing all of Radware’s proffered evidence together, the court cannot conclude that Radware has sufficiently .proven its entitlement to a conception date of [redacted].- To the extent that Radware’s proffered [redacted] A jury could reasonably conclude that Radware had not conceived of the invention by [redacted], and the court finds that there is a genuine issue of fact as to the conception date of the inventions disclosed in the ’319 and ’374 .patents.

b. Reduction to Practice

Radware asserts that the inventions disclosed in the ’319 and ’374 patents were reduced to practice at least as early as October 4, 1999, one day prior to the filing date of the Maki-Kullas patent. Nee-Dkt. No. 186-4 at 17-19. Although the court found that Radware has not proven its *990claimed conception date of [redacted], because an inventor cannot reduce to practice an invention of which he has not conceived, the court proceeds to examine whether Radware can demonstrate that the inventions disclosed in the ’319 and ’374 patents were reduced to practice by October 4, 1999. Proof that the inventions were reduced to practice by October 4, 1999 would also serve as proof that the invention was conceived prior to the filing date of the Maki-Kullas patent, and therefore that the ’319 and ’374 patents claim priority over the Maki-Kullas patent.

Radware offers the following evidence in support of its claimed reduction to practice date:

• Radware press release issued on or about October 4, 1999 announcing Link-Proof as “the first IP load balancing solution for networks with multiple ISP connections.” See Dkt. No. 189-6.

• Testimony of co-inventor Peles. [redacted]. Dkt. No. 186-9 at 84:14-18.10

To demonstrate actual reduction to practice,11 a party must show that the inventor: (1) constructed an embodiment or performed a process that met all the claimed limitations; and (2) determined that the invention would work for its intended purpose. Z4 Technologies, Inc. v. Microsoft Corp., 507 F.3d 1340, 1352 (Fed.Cir.2007) (citing Cooper v. Goldfarb, 154 F.3d 1321, 1327 (Fed.Cir.1998)); see also Estee Lauder Inc. v. L’Oréal, S.A., 129 F.3d 588, 592-95 (Fed.Cir.1997) (holding that reduction to practice does' not occur until inventor knows embodiment will work for its intended purposes). Like conception (and diligence in reducing an invention to practice), inventor testimony of reduction to practice must be corroborated. In re NTP, 654 F.3d 1279, 1291 (Fed.Cir.2011) (“An inventor cannot rely on uncorroborated testimony to establish a prior invention date____It has long been the case that an inventor’s allegations of earlier invention alone are insufficient — an alleged date of invention must be corroborated.) (citations omitted); Cooper, 154 F.3d at 1330 (“In order to establish an actual reduction to practice, an inventor’s testimony must be corroborated by independent evidence.”). Courts apply a rule of reason approach to the question of reduction to practice, but as with conception, “evidence of corroboration must not depend solely on the inventor himself.” Id. at 1330.

Radware argues that Peles’ testimony is corroborated by the LinkProof press release, which Radware argues is direct, independent, and contemporaneous evidence of the claimed October 4, 1999 date. Dkt. No. 209-4 at 6-7. However, Peles’ testimony is limited to the assertion that October 4, 1999 was the earliest date for which evidence of reduction to practice could be found. See Dkt. No. 186-9 at 84:14-18 [redacted] It does not appear that Peles affirmatively testified that October 4,1999 was the date by which Radware had reduced its invention to practice. The court therefore finds that his testimony does not constitute evidence of Radware’s claimed date for reduction to practice.

The question of whether Radware reduced the invention disclosed in the ’319 and ’374 patents to practice by October 4, *9911999 therefore largely hinges on the Link-Proof press release. Because the court finds that Radware has not established as undisputed that the press release was issued on October 4, 1999, the court finds the press release insufficient to demonstrate reduction to practice; Radware asserts in its motion for summary judgment that the press release was issued “on or about” October 4, 1999. Dkt. No. 185-4 at 3. While the press release itself includes the date “October 4, 1999,” the document produced by Radware also contains the following handwritten notation: “Issued Oct 1999.” Dkt. No. 189-6 at ECF p. 2. Radware produces no corroborating evidence that the press release was actually issued on October 4, 1999; In its reply, Radware argues that the document speaks for itself, and is consistent with the testimony of others who recall that the Link-Proof product was being offered for sale at least by October 4,1999; Dkt. No. 209-4 at 6-7. However, ' the testimony Radware cites does not necessarily support a publication date of October 4, 1999. Radware first cites the deposition testimony of Zack Cherkassy,12 who testified that [redacted] Dkt. 209-12 at 37:19. But Radware does not explain how ’ Cherkassy’s recollection that LinkProof was released in September or October, 1999 is consistent with its contention that LinkProof was announced on or around October 4, 1999, and the press release to that effect. Radware also cites Dr. Rubin’s supplemental declaration, in which he opines that [redacted] Dkt; No. 210-11, ¶ 6. The court fails to see how Dr. Rubin’s supplemental declaration helps establish that Radware issued the press release on October 4,1999.

• In sum, the court finds that a reasonable juror, based on the evidence provided by Radware, could conclude that the invention disclosed in the ’319 and ’374 patents was not reduced to practice-by October 4,1999. Although Radware has submitted some evidence that it reduced the invention to practice in October, 1999, there is a genuine issue of fact- as to whether it did so by October 4,1999.

Becausé the court finds that Radware has not conclusively shown that it either conceived of tbé invention disclosed in the ’319 and ’374 patents by April 1, 1999, or that Radware reduced its invention to practice by October 4, 1999, the court hereby DENIES Radware’s motion for summary judgment that the Maki-Kullas patent is not prior art as a matter of law.13

4. Cisco DistributedDirector White Paper

Finally, Radware moves for summary judgment that both versions of the Cisco DD white paper produced by F5 are not prior- art. Dkt. No. 189 at 19-22. Two versions of the of the DD white paper have been produced in discovery: (1) a version bearing a “1999” copyright notice which was authenticated by its author, Kevin Delgadillo (the .“’99 DD white paper”); and (2) a second version, which apparently has not been authenticated, and which bears a copyright date of 1997 (the “’97 DD white paper”). See Dkt. Nos. 189-15, 186-13. F5 asserts that both documents are anticipating prior art references and. support a finding that the ’319 and ’374 patents are invalid for obviousness.

The parties’ dispute regarding the ’99 DD white paper largely concerns whether *992F5 has shown that it was published prior to April 1, 1999 — the date by which Rad-ware contends the-inventions disclosed in the ’319 and ’374 patents were conceived. However, as discussed above, Radware has not shown that the Asserted. Patents are necessarily entitled to a priority date of April 1, 1999. F5 has introduced evidence that the ’99 DD white paper was released at least sometime in April of 1999,. and possibly in March of that year: (1) the document bears a “1999” copyright on every page, and bears the notation “3/99” on the final page, which Cisco employee and author of the document Kevin Delgadillo testified indicates the document was available publicly in March 1999, see Dkt. No. 189-18 at 17:19-18:22; and (2) Delgadillo testified that the document was published to the Cisco website in March 1999 (or, at the latest, April 1999), Dkt. No. at 3:16-18, 11:23-12:2, 13:11-24; and'16:3-15. Accordingly, the court finds that there is a genuine issue of fact as to the publication date of the ’99 DD white paper, and DENIES Radware’s motion for summary judgment that the ’99 DD white paper is not prior art as a matter of law. .

The second version, the ’97 DD white paper, bears a “1997” copyright on each page, but has not been authenticated by Delgadillo. This version is also “mutilated,” according to Radware, in that it has several blank spaces throughout. Dkt. No. 209-4 at 9. Radware contends that the court should’grant summary judgment that this version is not prior art because there is no evidence in the record that the document was ever publicly available. However, although it is currently unauthenticated and F5 chose not to depose Delgadillo about the document, Radware provides no reason to preclude F5 from-potentially introducing the document as evidence on the basis of Delgadillo’s- trial testimony. Moreover, Radware- concedes the ’99 and ’97 documents contain some of the same material, see Dkt. No. 186-4 at 22, which, along with the “1997” copyright date, at least raises an issue of fact as to the document’s authenticity and publication date. Accordingly, the court DENIES Radware’s motion for summary judgment that the ’97 DD white paper is not prior art as a matter of law.

C. Anticipation

• F5 moves for summary judgment of invalidity, arguing that all asserted claims of both the. ’319 and ’374 patents are invalid as anticipated by Cisco DD. Dkt. No. 183 at 14-20.14 For the reasons set' forth below, the court finds that F5.has not shown by clear and convincing evidence that Cisco DD anticipates the asserted claims, and that a reasonable jury could find that the Cisco DD system does not disclose ,ISP link load balancing. Accordingly, F5’s motion for summary judgment that the asserted claims are invalid as anticipated by Cisco DD is DENIED.

1. Legql Standard

A claim is anticipated under § 102, and thus invalid, “if each and every limitation is found either expressly or inherently in a single prior art reference.” Bristol-Myers Squibb Co. v. Ben Venue Labs., Inc., 246 F.3d 1368, 1374 (Fed.Cir.2001) (internal quotation marks and citation omitted); accord Eli Lilly & Co. v. Zenith Goldline Pharm., Inc., 471 F.3d 1369, 1375 (Fed.Cir.2006). Put simply, “[t]hat which infringes, if later, would anticipate, if earlier.” Peters v. Active Mfg., 129 U.S. 530, 537, 9 S.Ct. 389, 32 L.Ed. 738 *993(1889) (internal quotation mark omitted). To show anticipation' under 35 U.S.C. § 102, the moving party must “identify each claim element, state the witnesses’ interpretation of the claim element, and explain in detail how each claim element is disclosed in the prior ‘ art reference.” Schumer v. Lab. Computer Systems, Inc., 308 F.3d 1304, 1315-16 (Fed.Cir.2002). A reference can also anticipate “if that missing'characteristic is necessarily present, or inherent, in the single anticipating reference.” Schering Corp. v. Geneva Pharm., 339 F.3d 1373, 1377 (Fed.Cir.2003).

Anticipation under § 102 is a two-step inquiry. See Medichem, S.A. v. Rolabo, S.L., 353 F.3d 928, 933 (Fed.Cir.2003). The first step is-.claim construction. Id.; see Amazon.com, Inc. v. Barnesandnoble.com, Inc., 239 F.3d 1343, 1351 (Fed.Cir.2001) (“‘claim must’be construed before determining its validity just as it is first construed before deciding infringement.’ ” (quoting Markman v. Westview Instruments, Inc., 52 F.3d 967, 996 n. 7 (Fed.Cir.1995) (Mayer, J., concurring), aff'd, 517 U.S. 370, 116 S.Ct. 1384, 134 L.Ed.2d 577 (1996)). The second step is a comparison of the properly construed claim to the prior art. Medichem, 353 F.3d at 933.

2. Analysis

•The Court previously construed' the claims in its claim construction order. See Dkt. No. 122. As relevant here, the Court construed the following terms:

[[Image here]]

The asserted claims in this case comprise two types: claims that are directed to selecting a route for inbound messages from a remote computer to the multi-homed network (claims 1-23 and 29-32 of the ’319 patent and claims 1-4, 6-12, and 14-15 of the ’374 patent), and those directed to selecting one of multiple routes for outbound messages, where the client computer is located on a multi-homed network and is sending communications out to a remote server (claims 24-28 of the ’319 patent). See Dkt. No. 122 at 17-19. The inbound claims all contain a multi-homing limitation, a limitation for resolving a DNS query from a remote computer for a domain name within the multi-homed network, and, critically, a limitation for selecting one of a plurality of routes through One of the ISPs to áénd a response to the DNS *994query. Some inbound claims also include a limitation that the route selection be based on one or more metrics or criteria, such as latency, hops, costing, load, or data packet loss.15 Finally, some of the inbound claims contain a limitation requiring the translation of IP addresses for packets received on the multi-homed network to an IP address within that network, known as network address translation (“NAT” or “NATing”). Like the inbound claims, each outbound claim contains a multi-homing limitation and the limitation of selecting one of the plurality of routes. However, for the outbound claims, the client is located at the multi-homed network and is sending packets to a remote server. Additionally, the outbound claims include the limitation that route selection must be based on costing information of the routes and NATing. Although F5 contends that Cisco DD/ GSLB anticipates all asserted claims, the court discusses the inbound claims before turning to the outbound claims.

a. Inbound claims

F5 contends that Cisco DD anticipates all inbound claims of the ’319 and ’374 patents as it discloses each limitation required by the inbound claims. Radware disputes that Cisco DD meets the route selection limitation, and argues that the prior art does not disclose “route metrics” with regard to dependent claims, but otherwise does not dispute that Cisco DD discloses the other inbound claim limitations. See Dkt. No. 202-4.

i. Route Selection Limitation

Each claim requires the selection of either “one of a plurality of routes” connected to the internet (in the case of the ’319 patent) or the selection of “one ISP link from the plurality ISP links” (for the ’374 patent). The court previously construed ISP link to mean “a pathway connecting to or from an ISP.” Dkt. No. 122 at 20. The court stated that an “ISP link” is a subpart of a “route,” id. at 21, and the parties treat the route selection limitations of the two patents as functionally equivalent in the context of this motion.

In its reply, F5 concedes that GSLB did not select between different routes having the same destination, but rather between routes that had different destinations. Dkt. No. 211 at 6. (“Radware’s argument essentially boils down to the fact that prior art GSLB selected between routes had different destinations, not a single destination. This is true.... ”). F5 argues that this is a distinction without a difference: “[t]he concept of choosing a server based on well-known route metrics for GSLB is the same as choosing a route based on the same metrics for ISP link load balancing.” Dkt. No. 183 at 15. Furthermore, F5 argues that it has introduced direct evidence that Cisco DD was capable of comparing various routes to the same destination, and that it had been sold for that purpose. Id. According to F5, the result is that Cisco DD discloses the route selection limitation required by the asserted inbound claims.

F5’s evidence consists of the expert testimony of Dr. Alexander and his demonstration Cisco DD test-bed, and the witness testimony of Kevin Delgadillo and Dean Darwin. Dkt. No. 183 at 15. Dr. Alexander opined that “The [Cisco] DD System was capable of receiving a DNS resolution query from a remote computer for a domain name within the enterprise LAN computer network, selecting one of *995the plurality of routes connecting said device to the Internet, and responding to the DNS resolution query with an IP address associated with the selected route.” Dkt. No. 183-11 ¶ 173. According to Dr. Alexander, Cisco DD systems available as early as December 1998 “provided a plurality of routes for an enterprise Local Area Network (“LAN”) to connect to the Internet such that each of the plurality of routes could be assigned ISP specific IP addresses.” Id: ¶ 163. According to Dr. Alexander, Cisco DD can allocate each ISP “a virtual IP address from its pool of addresses so as to uniquely identify the respective ISPs.” Id. ¶ 183. Dr. Alexander explains that he was able to configure Cisco DD to perform the role of “a network controller receiving a DNS resolution query from a remote computer,” id. 1Í181, and to “respond.to the DNS query with one of a plurality of the virtual IP destination addresses associated with each of the respective ISP links.” Id. ¶ 184. Furthermore, his configuration of the Cisco DD system can “select, based on at least one load balancing criterion, one ISP link from the plurality ISP links.” Id. F5 further supports its argument with witness testimony. First,' Kevin Delgadillo testified that Cisco DD could accomplish ISP load balancing, and was likely used for that purpose. Dkt. No. 183-5 at 41:18-23. Second, Dean Darwin, a former Cisco employee, testified that he sold Cisco DD in 1997 and 1998 “to balance between [ ] ISP links” as an ISP link load balancing' solution. Dkt. No. 183-10 at 257:10-258:19.

Radware disagrees with F5’s analysis, and contends that GSLB, and by extension Cisco DD, “uses a routing protocol that does not consider multiple candidate routes to a destination for the purpose of providing load balancing among associated attached ISP links.” Dkt. 195-1 at' ¶6. According to Radware, this means that GSLB/Cisco DD chooses the best server, not necessarily the best route. Dkt. No. 202-4 at 17. Radware argues that the fundamental difference, between prior art GSLB systems like Cisco DD and inbound ISP LLB as claimed by the ’319 and ’3.74 patents is that inbound LLB does not make a selection between different destination servers; rather, the destination server remains the same, and the LLB system chooses the route (or ISP link) used to reach that server. Id. at 18. Furthermore, Radware’s expert takes issue with several aspects of the test-bed system, and contends that it fails to show that Cisco DD discloses the limitations of the asserted inbound claims.

The' court finds- that F5 has not shown by clear and convincing evidence that the Cisco DD system anticipates the asserted inbound claims. First, while Delgadillo does testify that Cisco DD could accomplish “ISP load balancing,” he also testified that he was “not aware of’ any Cisco products, including the Cisco Distributed-Director, that would “enable a company or enterprise to do link load balancing.” Dkt. No. 203-6 at 31:2-32:11. Delgadillo also testified that Cisco “never marketed Distributed Director as a link load balancing solution.” Id. at 41:25-42:10. In his testimony, Delgadillo distinguishes between “ISP load balancing” and “link load balancing,” a distinction which F5 does not explain. Second, Darwin testified that he sold Cisco DD for the purpose of providing ISP redundancy: “the ability to move traffic if... an ISP was failing or failing to provide the performance, if it was slow or if it was congested or it was overused, to be able to balance between the ISP links.” Dkt. No. 183-10 at 257:16-20. While his testimony may provide some support for F5’s position, he does not testify as to how Cisco DD “moved traffic” between links, and whether it did so by performing route selection akin to that taught by the Asserted Patents.

*996Finally, the parties have provided conflicting expert testimony on whether the test-bed system demonstrates the ability to perform inbound ISP link load balancing. While Dr. Alexander asserts that his test-bed system demonstrates Cisco DD is capable of performing ISP link load balancing, Dr. Rubin opines that even if all components of the Cisco DD System had pre-dated the priority date of the patents-in-suit,16 the Cisco DD System still would not anticipate because the system could not perform either inbound or outbound link load balancing. Dkt. No. 186-24 at ¶¶ 97-109.17 Both experts appear to present reasonable analyses of whether the Cisco DD system anticipates the asserted inbound claims software, and reach opposite conclusions. Summary judgment is therefore improper. Anderson, 477 U.S. at 248, 106 S.Ct. 2505; Celotex, 477 U.S. at 322-32, 106 S.Ct. 2548.

ii. Route Metrics Limitation

F5 also asserts that the inbound dependent claims are anticipated because such dependent claims require route metrics and such route metrics were known in the prior art and not invented by Radware. The inbound dependent claims all'require either that the route selection be: (1) “based on the proximity determination”; or more generally (2) “in accordance with one or more criteria of the plurality of routes.” See, e.g., ’319 col.19 11.35-36, 64-66. Although F5 argues that Cisco DD included the use of various metrics, including availability, proximity, and latency, Dkt. No. 183 at 18, Radware correctly notes that F5 fails to prove that such route metrics were used by a “device for managing a computer network, said device connected to the Internet through a plurality of routes,” wherein that device then “select[s] the one of the plurality of routes” for purposes of link load balancing in a multi-homed environment, Dkt. No. 202-4 at 19. For each type of criterion or metric required by the asserted claims, F5 simply notes that Cisco DD “includes” that criterion or metric, but does not “explain in detail how each claim element is disclosed in the prior "art reference.” Schumer, 308 F.3d at 1315-16.

b. Outbound Claims

The parties’ arguments regarding the outbound claims mirror those advanced for the inbound claims. F5 states .that like with the inbound claims, “as the outbound claims recite nothing more than these known GSLB techniques applied in the context of multi-homed networks, they are invalid.” Dkt. No. 183 at 20. Radware argues that “the outbound claims have a multi-homing route selection limitation that is not met by GSLB.” Dkt. No. 202-4 at 20.

The outbound claims' require route selection in a multi-homing network environment like the inbound claims, with the difference that the client, not the server, is located at the multi-homed network and is sending packets to a remote server, rather than a remote client. Accordingly, the court finds that for the reasons discussed above in connection with the inbound claims, F5 has not shown by clear and convincing evidence that Cisco DD anticipates the outbound claims. The outbound claims all also require the use of costing information, and for the same reasons stated in the court’s discussion of route metrics for inbound claims, the court finds that F5 has not explained in detail how Cisco DD discloses the use of costing information in the context of the asserted outbound claims.

*997In sum, the court finds that F5 has failed to carry its burden of showing by clear and convincing evidence that each limitation in the asserted claims is disclosed in a single prior art reference, Cisco DD. The court therefore DENIES F5’s motion for summary judgment .that the Asserted Patents are invalid as anticipated under 35 U.S.C. § 102.

D. Obviousness

The final issue to resolve is whether the asserted claims are obvious in light of the prior art, A patent is obvious if “the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains.” 35 U.S.C. § 103(a). See also KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 406, 127 S.Ct. 1727, 167 L.Ed.2d 705 (2007). What a particular reference discloses is a question of fact, see Para-Ordnance Manufacturing, Inc. v. SGS Importers International, Inc., 73 F.3d 1085, 1088 (Fed.Cir.1995), as is the question of whether there was a reason to combine certain references, see Transocean Offshore Deepwater Drilling, Inc. v. Maersk Contractors USA, Inc., 617 F.3d 1296, 1303 (Fed.Cir.2010). Under the four part test for obviousness detailed in Graham v. John Deere Co. of Kansas City, 383 U.S. 1, 17-18, 86 S.Ct. 684, 15 L.Ed.2d 545 (1966), the court must consider (1) the scope and content of the prior art; (2) the difference between the prior art and the claimed invention; (3) the level of ordinary skill in the art; and (4) any objective evidence of nonobviousness. The party asserting invalidity bears the burden of proving “by clear and convincing evidence that a skilled artisan would have been motivated to combine the teachings of the prior art references to achieve the claimed invention, and that the skilled artisan would have had a reasonable expectation of success in doing so.” Procter & Gamble Co. v. Teva Pharmaceuticals USA, Inc., 566 F.3d 989, 994 (Fed.Cir.2009) (quoting Pfizer, Inc. v. Apotex, Inc., 480 F.3d 1348, 1361 (Fed.Cir.2007)).

The scope and content of the prior art is set forth above. The parties agree on the level of ordinary skill in the art. See Dkt. No. 183 at 21 n.12. Radware has submitted evidence on objective indicia of nonobviousness.18 Although F5 relies on both GSLB/Cisco DD and BGP in support of its obviousness arguments, the court granted Radware’s motion to strike those portions of F5’s motion for summary judgment that rely on BGP. The court therefore does not consider F5’s arguments based on BGP.

F5 asserts that all' the core networking technology functionality of the claims existed in the prior art, and that to repur-pose those principles and apply them in multi-homing network use cases proposed by its customers, rather than in a geographically distributed server network, is nothing more than an application of well-known networking techniques that would have been obvious to a person of ordinary skill in the art. Dkt. No. 183 at 21. In support of its obviousness argument, F5 offers the testimony of Kevin Delgadillo and Dean Darwin that GSLB could potentially be repurposed for ISP link load balancing, as well as its assertion that Rad-ware “directly lifted source code from its GSLB product, WSD, for inclusion in its ISP link load balancing product.” Id. at 21-22. F5 does not offer any expert testimony regarding obviousness.

*998Radware’s expert disputes F5’s contention, opining that the “core networking functionality” that existed in the' prior art does not disclose, teach, or suggest ISP load balancing. Dkt. 195-1 ¶¶ 9-11. As discussed' above, Radware maintains that GSLB server load balancing is distinct from ISP load balancing. Dr. Rubin also opines that repurposing software modules is common in the software development industry, as such modules are universally shared through designs implemented by networking equipment, which has been done by F5 itself. Id. ¶ 12.

On this record, the court cannot say as a matter of law that it would have been obvious to one of ordinary skill in the art to repurpose the existing “core networking technology” for use in a multi-homing network environment. F5 bears the burden of showing by clear and convincing evidence that it would have been within .the skill of an ordinary artisan to do so, and F5 has not met that burden. The evidence F5 does present is contradicted by Dr.. Rubin’s expert opinion, and F5’s cursory assertion that it would be obvious to repurpose “core networking technologies” and apply them in a multi-homing network environment is insufficient to carry its burden on summary judgment. Innogenetics, N.V. v. Abbott Labs., 512 F.3d 1363, 1373 (Fed.Cir.2008) (a party must provide “articulated reasoning with some rational underpinning to support the legal conclusion of obviousness.”) (citations and quotation marks omitted). Additionally, the court finds that inconsistencies among the expert testimony introduced by Radware and F5’s evidence raise questions of credibility and weight. Therefore, issues of material fact remain with respect to obviousness. Accordingly, the court therefore DENIES F5’s motion for summary judgment that the Asserted Patents are invalid as obvious under 35 U.S.C. § 103.

IV. RADWARE’S MOTION FOR SANCTIONS UNDER RULE 37

Radware moves to exclude two categories of evidence under Federal Rule of Civil Procedure 37 based on F5’s alleged discovery violations: (1) evidence, especially source code, relating to- F5’s “hotfix” and (2) evidence relating to F5’s February 2015 review of its “QK view” files. For the reasons explained below, the court DENIES the motion-for sanctions.

A. Legal Standard

Under Rule 37(e)(1):

If a party fails to provide information or identify a witness as required by Rule 26(a) or (e), the party is not allowed to use that information or witness to supply evidence oh a motion, at a hearing, or at a trial, unless the failure was substantially justified or is harmless.

Rule 26(e) requires a party to supplement its disclosures and discovery responses if it learns of additional or corrective information. Rule 37(c)(1) “gives teeth to these requirements by forbidding the use” of any information required to be disclosed by Rule 26(a) or (e) that is not properly disclosed. Yeti by Molly Ltd. v. Deckers Outdoor Corp., 259 F.3d 1101, 1106 (9th Cir.2001). Rufe 37(c) provides a “self-executing, automatic sanction to provide a strong inducement for disclosure of material.” Id. at 1107; Hoffman v. Constr. Protective Servs., Inc., 541 F.3d 1175, 1179 (9th Cir.2008) (“Under Rule 37 exclusion of evidence not disclosed is appropriate unless the failure to disclose was substantially justified or harmless.”). “Two express exceptions ameliorate the harshness of Rule 37(c)(1): The information may be introduced if the parties’ failure to disclose the required information is substantially justified or harmless.” Yeti, 259 F.3d at 1106. It is F5’s burden to prove harmlessness or substantial justification. Id. at 1107.

*999The duty to disclose documents created after the close of discovery , is not clear. Tilton v. McGraw-Hill Companies, Inc., Case No. 06-0098, 2007 WL 3229157, at *2 (W.D.Wash. Oct. 30, 2007).

B. F5’s Hotfix

On November 4, 2014, F5 issued a “hotfix” to remove, certain functionality in the accused devices, . allegedly in order to ensure the accused devices did not infringe Radware’s Asserted Patents. Radware asserts that F5 did not promptly disclose the hotfix and did not produce the source code underlying the hotfix. Dkt. No. 207 at 3-4.

This court issued its claim construction order on April 18, 2014. Dkt. No. 122. On November 4, 2014,- Radware requested F5’s counsel to:. supplement the source code of the accused products to the. extent there were any material changes to the source code Radware had been provided to that point. Dkt. No. 207-2. Qn November 11, 2014 F5 responded that in lieu of producing the source code, Peter. Thornewell, F5’s 30(b)(6) witness on the technical operation of the source code, would testify as to any material changes at his. deposition. Radware deposed Peter Thornewell on November 12,2014, and he did not identify any material changes. On November 28, 2014, fact discovery closed. However, the parties stipulated that the deposition of certain witnesses, including Martin Brewer,. could take place after the close of fact discovery due to scheduling issues. Dkt. No. 225-3 at 3. On December 4, 2014, the day before Brewer’s deposition, F5 sent Radware an email stating that Brewer “will be prepared to give testimony regarding F5’s recent ‘End of Life announcement for inbound and outbound cost-based link load balancing and inbound link utilization-based load balancing,’” i.e. the hotfix. Dkt. No. 207-3. F5 also included a link to F5’s public announcement of the hotfix dated December 4,2014. Id.

On December 5, 2014, Brewer was deposed. See Dkt. Nos. 218-6, 218-7. Brewer testified about the how the BIG-IP accused products would function after the hotfix was implemented. Dkt. No. 218-7 at 240:20-241:21. He also testified to a limited extent as to how a customer' could receive a “patch” to re-enable the cost-based link load balancing but noted that F5 had not “quite figured through the details about who would be eligible and things like that.” Dkt. No. 198-10 at 64:13-15. Brewer also “[did] not know” whether the hotfix included “any other features or fixes” beyond removing the cost-based link load balancing functionality. Dkt. No. 206-6 at 52:8-13. Brewer did not produce a copy of the source code for the hotfix or patch. Rad-ware made no explicit request following Brewer’s depositión to examine the source code until it filed the current motion for sanctions. -

Radware served its opening expert report authoreci by Dr. Rubin on January 15, 2015. Dr, Rubin addressed the hotfix in his report, relying on Brewer’s testimony and F5’s public announcement. Dkt. No. 179-26 at ¶ 125, and n.56-58. Dr. Rubin was able to identify the specific functionalities that would be disabled by the hotfix: [redacted] Id. ¶ 129 (citing Brewer Dep.);. see also ¶ 158 [effect of the hotfix on inbound claims). F5 disclosed the report of its expert, Dr. Peter Alexander, on March 2, 2015. He also discussed the hotfix. See Dkt. No. 218-8 at 173-179 (Section 6.3 The F5 Hotfix Configuration). His testimony was general and based upon information from F5’s attorneys who presumably got the information from F5’s engineers. The experts appear to agree that the hotfix [redacted] Id. at ¶ 494-95, citing Rubin Rep. at ¶ 158.19

*1000On May 29, 2015 Mr. Thornewell" submitted a declaration in support of F5’s motion for. summary judgment of non-infringement stating that he “designed much of the code for the ‘hotfix,’ which; F5 implemented to disable certain functionalities relating to the ability of the-, products to distribute traffic across ISP links in a mul-ti-homed environment.” Dkt. No. 188-6 ¶ 3. Mr. Thornewell had not discussed the hot-fix at his November 12, 2014. deposition.

F5’s disclosure of the hotfix was purportedly made as soon as it became available: the day before Brewer’s deposition. Although the timing of the disclosure of the hotfix appears suspicious, there is no direct evidence that it was done to disadvantage Radware. F5 may not have produced the source code because it did not believe it would be necessary to use the source code itself to show non-infringement. Therefore, F5 may have assumed that Radware did not need or desire to review the hot fix source code. Nevertheless, to avoid any possible prejudice, the court orders that Radware be given the same opportunity to investigate the hotfix that it would have had if earlier disclosure had been made. The court therefore orders F5 to make the relevant source codé available for Radware’s review and to 'make Mr. Thornewell available for up to three hours of deposition on the design and implementation of the hotfix. Additionally, the court finds that any failure by F5 tó supplement its disclosures and discovery responses is rendered harmless by this order.

C. The QK View Files

In support of F5’s motion for summary judgment on damages issues, F5 refers to customer configuration data found in F5’s QK View files. The' QK View files are diagnostic snapshots of how a customer uses and configures F5’s BIG-IP products. Dkt. No. 218-3 at 3. Oscar Brain, an F5 witness, conducted three reviews of the QK View files: one in May 2014 and two in February 2015. Mr. Brain was deposed in October 2014 and testified about the May 2014 review, the contents of the QK View files, and how reports are génerated from the files. See Dkt. No. 218-4 at 3; Dkt. No. 225-5 at 2. Radware was not informed of the February 2015 reviews until F5 produced - Laura Stamm’s expert report on damages. ■

In her expert report, Ms. Stamm uses the February 2015 customér configuration reports to support her contention that there was no demand for the patentable feature of the accused LTM and GTM products'. Dkt. No. 184-14 ¶ 64. Radware was aware of the QK View database and the information available within the database during the discovery period, and deposed an F5 witness about the database. Radware also deposed Ms. Stamm about the- QK View files. See Dkt. No. 218-12. However, Radware has not had the opportunity to examine the data underlying Brain’s February 2015 configuration reports, which were created after the close of discovery. Accordingly, fairness dictates that F5 produce the data underlying the February 2015 reports on which Mr. Brain relied to create his February reports, and make Mr. Brain available for up to three hours of deposition testimony directed only at the basis and preparation of the February 2015 reports.

Radware’s motion for sanctions seeking exclusions of the source code for the hotfix and the February 2015 QK View reports is DENIED subject to F5’s production within thirty days or as otherwise agreed by the parties of: (1) the hotfix source code; (2) a deposition of Thornewell not to exceed three hours in length and limited to the design and implementation of the hot-*1001fix and patch; (3) the underlying data supporting the 2015 QK view reports; and (4) three hours of deposition from Mr. Brain concerning the-creation of the February 2015 QK View reports.

V. THE PARTIES’ MOTIONS FOR SUMMARY JUDGMENT OF INFRINGEMENT AND NON-INFRINGEMENT

Each party moves for summary judgment on infringement of various asserted claims. Radware moves for summary judgment of infringement on claims 1, 2, 7, 10, 11,12, and 24 of the ’319 Patent and claims 9, 10, and 12 of the ’374 Patent. Dkt. No. 182. For the reasons set forth below, Rad-ware’s motion is GRANTED for claim 24 of the ’319 Patent for pre-hotfix products but DENIED in all other respects.

F5 moves for summary judgment of non-infringement on five issues: (1) that F5 does not directly infringe any of the asserted method claims;20 (2). that F5’s Local Traffic Manager, standing alone, does not infringe any of the claims; (3) that the post-“hotfix” version of the LTM does not infringe any of the asserted claims; (4) that the post-“hotfix” version of the Link Controller does not practice claims 24-28 of the ’319 Patent; and (5) that the post-“hotfix” version of the Global Traffic Manager does not infringe claims 3-23, 29, 30, and 32 of the ’319 Patent and claims 5 and 13 of the ’374 Patent. Dkt. No. 190. The court GRANTS F5’s motion regarding: (1) no direct infringement of method claims; and (2) no infringement by LTM standing alone. As to the post-hotfix products, at the time the parties submitted their motions, F5 had not produced its post-hotfix source code to Radware; The court therefore DENIES WITHOUT PREJUDICE F5’s motion for summary judgment of non-infringement regarding post-hotfix products. Nevertheless, because the court will provide F5 with the opportunity to renew its motion once supplemental discovery is complete, the court discusses several arguments raised in F5’s motion below.

A. Legal Standard

Summary judgment on infringement is a two-step analysis. “First, the claims' of the patent must be construed to determine their scope. Second, a determination must be made as to whether the properly construed claims read on the accused device.” Pitney Bowes, Inc. v. Hewlett-Packard Co., 182 F.3d 1298, 1304 (Fed.Cir.1999) (internal citation omitted). “[S]ummary judgment of non-infringement can only be granted if, after viewing the alleged facts in the light most favorable to the .non-movant,■ there is no genuine issue whether the accused device is encompassed by the claims.” Id. at 1304.

B. Direct Infringement of Method Claims

F5 moves for summary judgment of no direct infringement of the. asserted method claims. Radware, “to conserve judicial resources and simplify the case for trial,... agrees to not- assert its theories of direct infringement for the method claims.” Dkt. No. 200-4 at 3 (Radware Opp. to F5. MSJ Nonlnf.). Accordingly, the court GRANTS F5’s motion of no direct infringement .of ,the asserted methods claims: ’319 Patent claims 13-23 and 26-28; and ’374 Patent claims 1-4 and 6-7.

C. Finjan and Fantasy Sports

Several of F5’s arguments on infringement depend on how Finjan, Inc. v. Secure Computing Corp., 626 F.3d 1197 *1002(Fed.Cir.2010), and Fantasy Sports Properties, Inc. v. Sportsline.com, Inc., 287 F.3d 1108 (Fed.Cir.2002), apply to the facts of this case. Having reviewed the applicable case law, the court concludes that F5’s accused BIG-IP products may infringe the Asserted Patents if the products include programs that carry out, or are configured to carry out, the claims of the Asserted Patents.21 As explained in more detail below, the BIG-IP products may be infringing because applications configured to carry out the claimed functions are preloaded onto the BIG-IP devices when sold, even though the applications must be enabled before use.

In Fantasy Sports the Federal Circuit found that a fantasy football program infringed claims directed to a system that awarded “bonus points.” 287 F.3d at 1117-19. The accused program was configurable to award bonus points, even though a user had to select various options already present in the software to get the bonus points. The Federal Circuit explained:

Although a user must activate the functions programmed into a piece of software by selecting those options, the user is only activating means that are already present in the underlying software. Otherwise, the user would be required to alter the code to enable the computer to carry out those functions. Accordingly, in order to infringe the ’603 patent, the code underlying an accused fantasy football game must be written in such a way as to enable a user of that software to utilize the function of awarding bonus points for unusual plays such as out-of-position scoring, without having to modify that code. In other words, an infringing software must include the “means for scoring.. .bonus points” regardless whether that means is activated or utilized in any way.

Id. at 1118 (emphasis in original).

In Finjan the Federal Circuit' evaluated claims of direct infringement by

three accused computer security products: a ‘Webwasher’ software download, a ‘Webwasher’ hardware ‘appliance’ or server containing software, and a ‘Cy-berguard TSP’ hardware appliance that also contains software. It is undisputed that all three products contain source code for eight software modules. Three of those modules — Anti-Virus, Anti-Mal-ware, and Content Protection — offer proactive scanning functionality, in addition to other features. The eight modules are ‘locked’ when the three products are sold, requiring a customer to purchase a separate key to activate each individual module. Therefore, a customer who purchases an accused product can activate all, some, or none of the eight modules at different cost.

626 F.3d at 1201-02 (citations omitted). The Federal Circuit first analyzed the specific claim language at issue:

Finjan’s non-method claims describe capabilities without requiring that any software components be “active” or “enabled.” The system claims recite software components with specific purposes: “a logical engine for preventing execution”, “a communications engine for obtaining a Downloadable”, or “a linking engine.. .for forming a sandbox package”. . .This language does not require that the program code be “active,” only that it be written “for causing” a server or a computer to perform certain steps.

Id. at 1204 (citations omitted).

Applying the analysis from Fantasy Sports, the Federal Circuit found that the *1003accused products, which contained the “software for performing the claimed functions” when sold, infringed the Asserted Patents. Id. at 1205. Specifically,

The code for proactive scanning was ‘already present’ in Defendants’ accused products when sold. There is no evidence that customers needed to modify the underlying code to unlock any software modules. The fact that users needed to ‘activate the functions programmed’ by purchasing keys does not detract from or somehow nullify the existence of the claimed structure in the accused software.

Id.

Here, the device claims, like those in Fantasy Sports and Finjan, describe the capability and configuration of the network controller. For example, claim 9 of the ’374 Patent requires “[a] device for load balancing client requests among a plurality of internet service provider (ISP) links in a multi-homed network, comprising a network controller configured to resolve an incoming domain name server (DNS) query.”

F5’s arguments attempting to distinguish Finjan and Fantasy Sports are not persuasive. F5 argues that claims 1 and 24 of the ’319 Patent and claim 9 of the ’374 Patent “all require active, enabled functionality” to be infringed. Dkt. No. 198-14 (F5 Opp. to Radware MSJ Inf.) at 6. F5 is correct that the court must look to the specific claim language to evaluate infringement. Here, the claim language is not distinguishable from that in Finjan. First, Finjan did not turn on special means-plus-funetion language. F5 Opp. to Radware MSJ Inf. at 8. Second, the court has already held that the claims do not require active method steps; rather, the court held that to infringe, an accused product must be delivered configured to perform the claimed functions. See Dkt. No. 145 at 6-9.

The court previously held that the device claims were not invalid for improperly mixing statutory classes. The court rejected F5’s argument that the functional language in the device claims called for an affirmative method step. Dkt. No. 145 at 6-9. The court found that “the claims at issue in this case recite a device or system with various components that perform specified functions. There is no confusion over when infringement occurs: “when one makes, uses, offers to sell, or sells the claimed apparatus: the [network controller.]” Id. at 9 (citing HTC Corp. v. IPCom GmbH & Co., KG, 667 F.3d 1270, 1277 (Fed.Cir.2012)). The court also rejected F5’s argument that “because the products as sold are not configured to perform the claimed functions” F5’s products did not infringe. Dkt. No. 145 at 17.

The cases that F5 cites in opposition, Telemac Cellular Corp. v. Topp Telecom, Inc., 247 F.3d 1316 (Fed.Cir.2001) and Typhoon Touch Techs., Inc. v. Dell, Inc., 659 F.3d 1376 (Fed.Cir.2011) are distinguishable from the instant case because the accused devices there were not actually configured to perform the claimed functions, but were only capable of being configured (i.e., programed) to perform the claimed functions. Telemac, 247 F.3d at 1330 (“[T]hat a device is capable of being modified to operate in an infringing manner is not sufficient, by itself, to support a finding of infringement. In this case, due to a restriction built into the software program stored in the telephone’s memory, a user of Topp’s system is prevented from directly placing international calls.”) (citation omitted); Typhoon, 659 F.3d at 1380 (“The district court construed the claim as requiring that a device, to be covered by the claim, actually performs, or is configured or programmed to perform, each of the functions stated in the claim,” in contrast to merely having “the capability of being configured or programmed to per*1004form the stated function, although not so structured in the device provided by a defendant.”).

The court’s construction of “configured” captures this distinction. The court previously construed the term “configured to” to mean “ ‘programmed to [perform certain functions.]’ This does not require user intervention if the feature claimed is included in the product as supplied.” Dkt. No. 122 at 21. The court noted:

In general, the court agrees with Rad-ware that “configured to” does not require user activation, but does require “that the claimed feature be included in the software.” Dkt. No. 136- (Radware ■Br.) at 13 (citing Fantasy Sports Properties, Inc. v. Sportsline.com, Inc., 287 F.3d 1108, 1118 (Fed.Cir.2002)). Thus, merely being “capable of’ performing a function is not enough, but if a device comes programmed with specific claimed functions it falls within the claims.

Id. at 22.

Thus, like in Fantasy Sports, F5’s products- infringe if they contain code “written in such a way as to enable a user of that software to utilize the [claimed] function.. .without having to modify that code.” 287 F.3d at 1118. What cases like Fantasy Sports, Finjan, and Typhoon Tech, recognize is that nearly all computers can be programed to perform claimed software functions. A- computer with a mémory, keyboard, display,' and processor is “capable” of being “configured” to carry out, -infinite software functions. However, selling such a computer does not constitute infringement if the purchaser is the one who has to program the computer to carry out the claimed functions. If the infringer, on the other- hand, sells a computer preloaded with infringing software code, the computer is “configured,” or “operable,” to perform the claims, and may be infringing under Finjan and Fantasy Sports. Cf. Nazomi Comm., Inc. v. Nokia Corp., 739 F. 3d 1339, 1346 (Fed.Cir.2014) (“[T]he structure (i.e., JTEK software)- necessary to enable Jazelle hardware to process stack-based instructions (i.e., Java bytecodes) is not only inactive, it is not -even present on the accused products. The installation of JTEK software is not unlocking existing functionality, but adding new functionality not currently present. There is no infringement.”)-

Here, F5 does not dispute that the BIG-IP products are sold and delivered to customer with [redacted] preloaded, although not accessible until licensed. See Dkt. No. 200-4 (Radware Opp. to F5 MSJ Nonlnf.) at 4. Accordingly, if the software preloaded onto the BlG-IP products is configured to perform the claimed functions, the BIG-IP products may be infringing. The court now turns to the parties’ arguments regarding infringement of specific claims.

D. Outbound Link Load Balancing: ’319 Patent claims 24-28

Independent claim 24 of the ’319 Patent recites (emphasis added):

24. A routing device for routing data via a network from a first node to a second node,' said network having a plurality of available routes from said first node to said second node and -the plurality of routes are assigned with respective IP addresses, said routing device comprising: '

a route selector operable to select one of said routes for sending data between said first node and said second node on the basis of costing information of said respective routes', and

a network address translator operable to receives a packet having a source IP address and translating the source IP address to an IP address corresponding to the selected route of the plurality of routes.

*1005The court construed “costing information” as having its “plain and ordinary-meaning, not limited to monetary price.” Dkt. No. 122 at 25-26.

1. Pre-hotfix: ’319 Patent Claim 24

Radware moves for summary judgment of infringement of claim 24 of the ’319 Patent. F5’s non-infringement argument is-that the Local Traffic Manager (“LTM”) “standing alone” is not capable of infringing the “costing information” element. Dkt. No. 190 (F5 MSJ Nonlnf.) at 8. F5 notes that Radware’s expert cites to Link Controller and. GTM functionality to support his infringement opinion. Id. At .the hearing on this motion, F5 emphasized that whether or not LTM “alone” infringes the Asserted Patents was relevant to the parties’ damages analysis. F5 specifically sought a determination from the court that LTM alone does' not infringe. The court notes that whatever effect LTM versus LTM plus other modules has on the parties’ damages opinions is not relevant to the infringement question. In any event, Radware has not shown that LTM “alone” infringes the Asserted Patents. All of the outbound claims at issue require load balancing based on cost. The documents that Radware points to suggesting th.e LTM can load balance do not support a finding that LTM can load balance based on cost. Dkt. No. 200-3 at 7-8. For example, the “BIG-IP Local Traffic Manager Implementations” document only states that, after following various set-up steps, “the virtual server is configured to load balance outbound connections to the routers.” Id. (citing Dkt. No. 182-21 at Radwar-eF500791322); This' says nothing -about load balancing based on cost. Accordingly, the court finds that LTM “alone” does not infringe claims 24-28 of the ’319 Patent.

Nonetheless, the BIG-IP accused products, pre-hotfix, may infringe claims 24-28 of the ’319 Patent.22 The fact that LTM “alone” does not infringe is not dispositive of whether the BIG-IP products infringe under Finjan because F5 admits that the pre-hotfix BIG-IP products were loaded with software that is configured to perform the functions claimed in the ’319 Patent. See Dkt. No. 188-6 (Thornewell Decl.) at ¶ 8 (“LTM has never been able, by itself, to load balance outgoing messages across ISPs based on cost or costing information, or any route-based metrics, without being paired with a GTM or Link Controller”). Accordingly, as F5 does not dispute that the pre-hotfix applications preloaded onto the BIG-IP product, considered together, infringe the outbound claims, the court GRANTS Radware’s motion for summary judgment of infringement as to claim 24 of the ’319 Patent by pre-hotfix products.

2. Post-hotfix: ’319 Patent Claims 24-28

As noted above, to allow for development of the factual record, the court DENIES WITHOUT PREJUDICE F5’s motion for summary judgment that F5’s post-hotfix products do not infringe. However, were the court to consider F5’s motion with respect to claims 24-28 on the present record, material disputes of fact would preclude summary judgment.

Based on F5’s representations of how its post-hotfix products'operate, Radware relies on the “Least Connections, Observed, Predictive and Bandwidth” metrics to show infringement of the “costing” limitations of claims 24-28. Dkt No. 200-3 at 8-10. The parties do not dispute that Least Connections is a measure of the number of connections at a particular node. Dkt. No. 190 at 11. Both Dr. Rubin and Dr. Alexander, the parties’ infringement experts, describe Least Connections as a “relatively simple [method] in that the LTM system *1006passes a new connection to the node that ■has the least number of current connections.” Alexander Decl. ¶ 485 (citing Rubin).

F5 argues that Least Connections does not measure costing information of the “entire route,” as required by the claims. Id. The court agrees that Least Connections is not a measure of “costing information of said respective routes,” and therefore does not infringe claims 24-28 of the ’319 Patent.

As the court explained in its prior summary judgment order, a measurement based on “said respective routes” must be taken over the entire route. Dkt. No. 145 at 21-22. In addressing AlO’s motion for summary judgment of non-infringement on AlO’s AX Series product, the court found that the AX Series, which only measured “the properties of immediately adjacent routers” did not meet the claim limitation of “ratings based on proximity' measurement taken through each of the plurality of routes.” Id. The court reasoned that “[t]he ’702 Patent contemplates taking proximity measurements through at least the entire route to the remote server ... Radware cannot point to any part of the ’702 Patent that discloses polling requests that are directed only at an adjacent router as is found in the AX Series products.” Id. The court reaches the same conclusion as to the costing claims of the ’319 Patent.23

The claim language as a whole further supports this conclusion. Claim 24 of the ’319 Patent first defines a route as extending “from said first node to said second node” and then specifies that the route selector selects an ISP based on “costing information of said respective routes.” The court also construed the phrase “a plurality of available routes from said first node to said second node” as “a. plurality of available routes connecting said first node to said second node through the Internet,” again showing that the route extends from the client computer to the remote server. Dkt. No. 122 at 19-20. By defining a “route” as extending from the first node to the second node, the claim clearly contemplates the entire route.

Here, the Least Connections metric does not measure a “cost” of the route — it only measures load on one portion of the route.24 Because the Least Connections metric is not representative of “costing information of said respective routes,” where the “route” must “conneet[] said first node to said second node through the Internet,” Least Connections does not infringe claims 24-28 of the ’319 Patent. To show infringement, Radware would need to rely on a costing metric other than “Least Connections,” “Observed,” or “Predicted.”

In fact, Radware does rely on another metric to prove infringement, the alleged “Bandwidth” method of route selection. See Dkt. No. 200-3 at 8 (citing Rubin Rep., Dkt.' No. 182-34 ¶¶126, 128-29). F5 acknowledges that some of its documentation describes a bandwidth metric but states that no such metric exists in actual Rad-ware products. Dkt. No. 190 at 11 (citing Alexander and Thornewell). F5 further notes that Radware’s expert fails to identify any F5 code — even pre-hotfix — implementing the supposed bandwidth metric. Id. The court finds that there is a material dispute of fact as to whether F5’s BIG-IP product performs a “bandwidth” method of *1007route selection. The court therefore DENIES F5’s motion for summary judgment of non-infringement as to claims 24-28 of the ’319 Patent as to post-hotfix products.

E. Inbound Link Load Balancing

Radware asserts infringement of claims 1-7, 9-19, 21-23, and 29-32 of the ’319 Patent and claims 1-4, 6-12, 14, and 15 of the ’374 Patent. Radware moves for summary judgment of infringement on claims 1, 2, 7,10,11 and 12 of the ’319 Patent and claims 9, 10, and 12 of the ’374 Patent. F5 moves for summary judgment of non-infringement on the method claims (addressed above), on claims 3-12, 29, 30, and 32 of the ’319 Patent, and claims 5 and 13 of the ’374 Patent.

The court'groups the device claims together as follows:

• Claims 1, 2, and 12 of the ’319 Patent, and claims 9, -10, and 12 of the ’374 Patent, which do not claim selecting a route based on a measurement “of said respective route”;

• Claims 3-5, 29, 30 and 32 of the ’319 Patent and claims 5 and 13 of the ’374 Patent, which claim proximity measurements;

• Claims 6-9 and 11 of the ’319 Patent, which relate to selecting a route based on a measurement “of said respective route”; and •

• Claim 10 of the ’319 Patent, which requires an Internet connection.

1. Claims 1, 2, and 12 of the ’319 Patent and Claims 9, 10 and 12 of the ’374 Patent

Radware’s expert Dr. Rubin opines that the BIG-IP products infringe claims 1, 2, and 12 of the ’319 Patent, and claims 9, 10, and 12 of the ’374 Patent. See Dkt. No. 179-26 (Rubin Rep.) at ¶¶ ISO-152, 154-157, 162-164, 188, 189, 192, 193, 209, 210, 220, 221, 224, 225, 228, and 229. In response, F5’s expert Dr. Alexander opines that the GTM and Link Controller do not infringe independent claim 1 of the ’319 Patent or independent claim 9 of the ’374 Patent. Dkt. No. 198-12 (Alexander Rep.) at ¶¶ 41-72, 308-341. Dr. Alexander opines that Dr. Rubin is relying on global server load balancing (GSLB) functionality, and not link load balancing functionality, to support his infringement opinions. Id.

Radware argues that Dr. Alexander’s report is not sufficient to rebut Dr. Rubin’s report because Dr. Alexander did not perform an infringement analysis. However, Dr. Alexander did “review Professor Ruben’s [sic] report and provide a rebuttal” to Dr. Rubin’s infringement analysis. Dkt. No. 210-7 (Alexander Dep.) at 12:13-14. Radware’s concerns about the level of detail in Dr. Alexander's' report do not address Dr. Alexander’s opinion that Dr. Rubin only identified GSLB functionality, and not link load balancing, in his infringement opinions. See Radware Reply at 10-11 (criticizing Dr. Alexander’s review of the hotfix). The two expert reports both appear to present reasonable analyses of the BIG-IP software, and reach different conclusions. Summary judgment is therefore improper, Anderson, 477 U.S. at 248, 106 S.Ct. 2505; Celotex, 477 U.S. at 322-32, 106 S.Ct. 2548.

Accordingly; the court finds that there is a material dispute of fact as to whether F5’s BIG-IP product is configured to perform inbound link load balancing, as claimed in claims 1, 2, and 12 of the ’319 Patent, and claims 9,10, and 12 of the ’374 Patent. The court therefore DENIES the motions for summary judgment as to claims 1, 2, and 12 of the ’319 Pátent, ahd claims 9, 10, and 12 of the ’374 Patent.

2. Claims 6-9 and 11 of the ’319 Patent

■ Claims 6-9 and 11 of the ’319 Patent require selecting an ISP “in accordance with one or more criteria of the plurality of *1008routes.” ’319 Patent cl. 11. As explained above, the court finds that a criteria “of the route” means a measurement taken over the entire route. Radware cannot rely on Least Connections or similar measurements that relate only to a portion of the route to show infringement. Accordingly, Radware’s motion for summary judgment of infringement regarding claims 7 and 11 is DENIED. F5’s motion for summary judgment of non-infringement of claims 6-9 and 11 of the ’319 Patent by post-hotfix products is DENIED WITHOUT PREJUDICE fop the reasons described above.

3. ’319 Patent Claims 3-5, 15-17, 29, 30, and 32 and ’374 Patent Claims 5 and 13: Proximity

F5 argues that the post-hotfix GTM does not measure any “proximities” through the ISP routes. Dkt, No. 190 (F5 MSJ Nonlnf.) at 16-17. Claims 3-5, 15-17, 29, 30, and 32 of the ’319 Patent require, for example, that the network controller determines the “proximities of remote computers to the computer network via the plurality of routes and selects one of the plurality of routes based on the proximity determination.” ’319 Patent cl. 3. The court construed “proximity” to mean “A measurement or measurements based on hops, latency,' TTL, or a combination thereof.” Dkt. No. 122 at 27. As discussed above, if F5’s factual representations about its post-hotfix products are true, the post-hotfix Link Controller,. GTM, and LTM can [redacted] as well as [redacted]. Rubin Report ¶ 158. However, none of .these metrics are “proximity” measures. Radware does not offer any counter to this conclusion. See Dkt. No. 200-4 (Radware Opp. to F5 MSJ Nonlnf.) Nevertheless, as discussed above, the court is unwilling to grant summary judgment based on an incomplete factual record with respect to post-hotfix products. Accordingly, the court DENIES WITHOUT PREJUDICE F5’s motion for summary judgment of-non-infringement that F.5’s post-hotfix products do not infringe claims 3-5, 15-17, 29, 30, and 32 of the ’319 Patent and claims 5 and 13 of the ’374 Patent.

4. ’319 Patent claim 10

Claim 10 of the ’319 Patent requires that. the link load balancing device be connected to the Internet:

1. A device for managing a computer network, said device connected to the Internet through a plurality of routes, wherein the plurality of routes are assigned with respective IP addresses, comprising:,.

10. The device of claim 1, wherein said device is connected within the computer network and is further connected to the Internet through the plurality of routes.

In the prior motions for summary judgment, F5 -moved for summary judgment of non-infringement, arguing that the claims of the.’319 Patent “require, among other things, that the claimed apparatus be connected to the Internet through a plurality of routes.” Dkt. No. 91 at 7 (F5 3/7/14 MSJ). F5 did not specifically address claim 10 of the ’.319 Patent, but focused on the preamble of certain claims in the ’702 Patent. Id. In denying F5’s motion, the court found that the preamble of Claim 1 of the ’319 Patent “does not contemplate that the Internet and multiple[] Internet connections are structural components of the device,” Dkt. No. 145 at 16. F5 now specifically addresses claim 10, and argues that this claim affirmatively requires that the device is connected to the Internet. Dkt. No. 198-14 (F5 Opp. to Radware Inf. MSJ) at 2, The court agrees. Claim 10 requires that the device be connected to. the Internet in the body of claim, requiring more than mere capability and describing more than just the “purpose or intended use” of the device. Rowe v. Dror, 112 F.3d 473, 478 (Fed.Cir.1997); see also MPEP § 2111.02 (9th ed. Mar. 2014). As there is no dispute *1009that F5’s device are sold unconnected to the Internet, the court finds that F5 does not directly infringe Claim 10 of the ’319 Patent by making or selling its BIG-IP products. Accordingly, the court GRANTS F5’s motion for summary judgment of non-infringement as to Claim 10 of the ’319 Patent. This applies to direct infringement of both pre- and post-hotfix versions of BIG-IP.25

VI. F5’S MOTION FOR SUMMARY JUDGMENT ON DAMAGES ISSUES

F5 moves for summary judgment on three damages issues: (1) Radware is not entitled to pre-filing damages; (2) there is no willfulness as a matter of law; and (3) Radware is not entitled to recover lost profits on the accused LTM and GTM products. Dkt. No. 187 (F5 Damages MSJ). The Court GRANTS the motion for summary judgment as to pre-filing damages and lost profits, and DENIES the motion for summary judgment as 'to willfulness.

A. Damages for Activities Prior to Filing of Complaint

F5 argues that Radware is not entitled to damages incurred prior to the filing of the complaint because Radware did not provide actual or constructive notice of infringement under 35 U.S.C. § 287.

1. Legal Standard

In order to recover damages for infringing activities prior to the filing of the complaint, a patentee must comply with 35 U.S.C. § 287(a):

Patentees, and persons making, offering for sale, or selling within the United States any. patented article for or under them, or importing any patented article into the United States, may give notice to the public that the same is patented, either by fixing thereon the word “patent” or the abbreviation “pat”, together with the number of the patent, or by fixing thereon the word “patent” or the abbreviation “pat.” together with an address of a posting on the Internet, accessible to the public without charge for accessing thé address, that associates the patented article- with the number of the patent, or when, from the character of the article, this cari not be done, by fixing to it, or to the package wherein one or more of thém is contained, a label containing a like notice. In the event of failure so to mark, no damages shall be recovered by the patentee in any action for infringement, except on proof that the infringer whs notified of the infringement and continued to infringe thereafter, in which event damages may be recovered only for infringement occurring after such notice. Filing of an action for infringement shall constitute such notice.

“Thus, [§ 287(a)] defines that ‘[a paten-tee] is entitled to damages from the time when it either began marking its product in compliance with, section ■ 287(a)[, constructive notice,] or when it actually notified [the accused infringer] of its infringement,-whichever was earlier.’ ” Maxwell v. J. Baker, Inc., 86 F.3d 1098, 1111 (Fed.Cir.1996) (quoting American Medical Sys., Inc. v. Medical Eng’g Corp., 6 F.3d 1523, 1537 (Fed.Cir.1993)) (formatting :in original), In Texas Digital Systems, Inc. v. Telegenix, Inc., the Federal Circuit noted that 35 U.S.C. § 287(a) “limits the extent to which damages may be recovered where products covered by a U.S. patent are sold *1010without the notice defined in the statute.” 308 F.3d 1193, 1220 (Fed.Cir.2002).

2. Analysis

F5 argues that Radware is not entitled to damages incurred prior to the filing of the complaint because Radware did not provide actual or constructive notice of infringement under § 287. The parties do not dispute that Radware did not provide F5 with actual notice of infringement until the filing of the complaint. F5 argues that Radware also failed to provide constructive notice by selling unmarked products in the U.S. prior to the filing of the complaint. Radware contends that by marking some, but not all of its products, Radware provided constructive notice.

Radware’s AppDirector, Alteon, and LinkProof products practice one or more of the claims of the Asserted Patents. Specifically, the AppDirector and Alteon products practice the asserted claims of the ’319 and ’374 Patents. See Dkt. Nos. 184-8 (Trachtman Dep.) at 26-29; 185-1 (Peles Dep.) at 79-80. The LinkProof product practices the asserted claims of the ’319 Patent. See Dkt. No. 179-26 (Rubin Rpt.) ¶ 91. Radware began marking the LinkProof on December 5, 2012, but did not begin marking the AppDirector and Alteon until August 6, 2013, several months after the filing of the complaint. See Dkt. No. 184-8 (Trachtman Dep.) at 186-188. Radware therefore sold unmarked AppDirector and Alteon products in the United States prior to filing its complaint. See, e.g., Dkt. Nos. 215-7 [redacted]; 215-8 [redacted]26

Radware argues that this limited marking was justified [redacted]. See Dkt. Nos. 204-7 (Zisapel 02/20/2014 Dep.) at 243:2-10; 204-9 (Trachtman Dep.) at 163:14-165:10. Thus, customers who purchased the unmarked products would have been directed to purchase a marked product, thereby placing the public on notice of the ’319 Patent as of December 5, 2012. Dkt. No. 204-4 (Radware Damages Opp.) at 4. Radware argues that this “virtual marking of the LinkProof’ constitutes sufficient constructive notice. See Radware Damages Opp. at 5. However, virtual marking of a product is not “substantially consistent and continuous” to constitute constructive notice. See Nike, Inc. v. Wal-Mart Stores, Inc., 138 F.3d 1437, 1446-47 (Fed.Cir.1998) (remanding case to the district court for a factual determination regarding whether the patentee complied with the marking statute and when compliance was achieved). A patentee must comply with § 287(a) to “obtain the benefit of constructive notice.” Maxwell v. J. Baker, Inc., 86 F.3d 1098, 1111 (Fed.Cir.1996) (finding that patentee complied with marking statute by diligently attempting to ensure proper marking by licensee).

Here, it was Radware’s responsibility to supply constructive notice of infringement via “substantially consistent and continuous” the marking of the products that practice the Asserted Patents. Nike, 138 F.3d at 1446-47. Merely encouraging a customer to buy a marked LinkProof product in combination with the unmarked AppDirector and Alteon products is not sufficient to constitute constructive notice under § 287(a). Accordingly, the court finds that Radware did not supply actual or constructive notice of infringement to F5, as required under § 287(a), and therefore Radware is not entitled to pre-suit damages. The court therefore GRANTS *1011F5’s motion for summary judgment as to pre-suit damages.

B. Increased Damages for Willfulness

F5 moves for summary judgment that Radware is not entitled to enhanced damages for willfulness as a matter of law. As discussed above, Radware is not .entitled to pre-suit damages. However, F5’s conduct prior to the filing of the complaint is still relevant to evaluating F5’s post-complaint conduct. As described below, Radware presents a material dispute of fact as to whether F5 was aware of the ’319 Patent and engaged in conduct relevant to willfulness prior to the filing of the complaint. Therefore, the rule laid out in In re Seagate Tech, LLC, 497 F.3d 1360, 1374 (Fed.Cir.2007), which prohibits enhanced damages “based solely on post-filing conduct” absent filing a preliminary injunction, does not necessarily bar willfulness in this case. Accordingly, the court DENIES the motion for summary judgment as to willfulness.

1. Legal Standard

The Patent Act requires a court to impose damages “adequate to compensate for the infringement.” 35 U.S.C. § 284. The Act also allows enhanced damages for willful infringement. Id.; King Instruments Corp. v. Perego, 65 F.3d 941, 947 (Fed.Cir.1995). To establish willfulness, “a patentee must show by clear and convincing evidence that (1) the infringer acted despite an objectively high likelihood that its actions constituted infringement of a valid patent,” and (2) the objectively-defined risk “was either known or so obvious that it should have been known to the accused infringer.” Seagate, 497 F.3d at 1374.

However, “[k]nowledge of a patent does not mean willfulness.” Honeywell Int'l Inc. v. Universal Avionics Sys. Corp., 585 F.Supp.2d 636, 644 (D.Del.2008); see also Norian Corp. v. Stryker Corp., 363 F.3d 1321, 1332 (Fed.Cir.2004) (“Willful infringement is not established by the simple fact of infringement, even though [the accused infringer] stipulated that it had knowledge of the [patentee’s] patents.”). The Federal Circuit has stated:

[A] willfulness claim asserted in the original complaint must necessarily be grounded exclusively in the accused in-fringer’s pre-filing conduct. By contrast, when an accused infringer’s post-filing conduct is reckless, a patentee can move for a preliminary injunction, which generally provides an adequate remedy for combating post-filing' willful infringement. A patentee who does not attempt to" stop an accused infringer’s activities in this manner should not be allowed to accrue enhanced damages based 'solely on the infringer’s post-filing cpnduct.

In re Seagate, 497 F.3d at 1374.

Furthermore, “[t]he ‘objective’ prong of Seagate tends not to be met where an accused, infringer relies on a reasonable defense to a charge of infringement.” Spine Solutions, Inc. v. Medtronic Sofamor Danek USA, Inc., 620 F.3d 1305, 1319 (Fed.Cir.2010); see also Innovention Toys, LLC v. MGA Entm’t, Inc., 611 Fed.Appx. 693, 700 (Fed.Cir.2015) (“[W]illfulness is not established where the defendant has a substantial.. .defense.”). Objectively reasonable defenses may include questions of infringement and validity. Bard Peripheral Vascular, Inc. v. W.L. Gore & Assocs., 682 F.3d 1003, 1006 (Fed.Cir.2012).

2. Analysis

F5 argues' that Radware is not entitled to seek increased damages for willful infringement of the ’319 Patent because: (1) F5 was unaware of the ’319 Patent prior to the filing of the complaint; (2) Radware did not seek a preliminary injunction; and (3) F5 has objectively reasonable defenses as a matter of law.

*1012First, F5 argues that Radware is not entitled to enhanced damages for willfulness because F5 was not aware of the ’319 Patent prior to the filing of the complaint. Radware argues ■ that F5 had notice ■ of the ’319 Patent because: (1) F5 had notice of the related ’702 Patent; (2) F5 cited the ’319 Patent Application in its own patent applications; and (3) the USPTO issued a Notice of Allowance to F5 that cited the ’319 Patent. Radware Damages Opp. at 8-10. Specifically, Radware points to evidence suggesting that F5 had notice of the related ’702 Patent as of January 2004, and cited it in connection with F5’s own patent applications. See Dkt. Nos. 204-24 (F5 internal email containing Radware announcement); 204-19 (F5 Answers to Rad-ware Interrogatories) at 21-22. The ’319 Patent was also cited in a Notice of Allowance sent from the USPTO to F5’s representative concerning an F5 patent, where it was listed under the category “Notice of References Cited.” Id. Finally, F5 also cited the ’319 Patent Application in the course of'prosecuting its own patents. See, e.g., Dkt. Nos. 204-10 (Skene Depo Ex.); 204-15 (F5 email attaching F5’s U.S. Patent Application 12/259,142).

F5 argues that its notice of the related ’702 Patent and ’319 Patent Application does not constitute notice of the ’319 Patent and that even if F5 did indeed receive the Notice of Allowance citing thé ’3Í9 Patent from its outside counsel, “[tjhere is no reason to think that F5 (or even its outside counsel) would have reviewed the patent.” Dkt. No. 215-3, F5 Damages Reply at 3, 5.

Radware’s evidence creates a material dispute of fact as to whether F5' knew about the ’319 Patent prior to the filing of the complaint. Both parties cite SoftView LLC v. Apple Inc. in support of their respective arguments. Case No. 10-389, 2012 WL 3061027, at *5-7 (D.Del. July 26, 2012). In that case, the court granted defendant Kyocera’s motion to dismiss willfulness allegations, finding that SoftView’s allegations of pre-suit knowledge due to media publicity did not support a reasonable inference that Kyocera was aware of the patent. Id. at *6. However, the court did not deny the willfulness allegations against AT&T, finding that there was a “plausible basis from which one might reasonably infer that AT&T had knowledge of the patent-in-suit prior to his litigation.” Id. The court considered three separate bases for allegations of AT&T’s pre-suit knowledge, including AT&T citing a published parent application of the patent in suit while prosecuting one if its own patents. Id. at *5. Similarly, here, F5 cited the ’702 Patent and the ’319 Patent Application while prosecuting its own patents. See, e.g., Dkt. Nos. 204-15 (F5 email attaching F5’s U.S. Patent Application 12/259,142); 204-19 (F5 Answers to Rad-ware Interrogatories) at 21-22.

Second, F5 argues that it is entitled to summary judgment on willfulness because Radware did not seek a preliminary injunction. F5 Damages MSJ at 2-3. Rad-ware disagrees. Radware Damages Opp. at 11-12. Both parties rely on provided language by the Federal Circuit in In re Seagate to support their positions. The pertinent language in In re Seagate states that a patentee who does not seek a preliminary injunction “should not be allowed to accrue enhanced damages based solely on the infringer’s post-filing conduct.” 497 F.3d at 1374 (emphasis added); see also SoftView, 2012 WL 3061027 at *8 (denying motion to dismiss plaintiffs willfulness claim, where plaintiff did not move for a preliminary injunction because plaintiff adequately alleged pre-suit and post-suit conduct). Here, as discussed above, Radware’s evidence creates a material dispute of fact as to whether F5 had pre-suit knowledge of the ’319 Patent.

*1013Moreover, the Federal Circuit has said that a "willfulness analysis should consider the “totality of the surrounding circumstances.” Graco, Inc. v. Binks Mfg. Co., 60 F.3d 785, 792 (Fed.Cir.1995). Radware raises some evidence of F5’s pre-filing conduct that could support willfulness. Hence, Radware’s willful infringement claims are not based solely on F5’s post-filing conduct, but are also based on allegations of F5’s pre-filing conduct concerning the ’319 Patent.

Specifically, Radware alleges that F5’s pre-filing conduct shows an awareness of the’319 Patent coupled with a desire to enter the field of link load balancing. See, e.g., Dkt. Nos. 179-7 (F5 11/20/2000 email chain) at 029107-08; 179-9 (F5 10/2001 document) at 039434, 039442; 204-15 (F5 email attaching, F5’s U.S. Patent Application 12/259,142); 204-19 (F5 Answers to Rad-ware Interrogatories) at 21-22. A jury could find evidence of F5’s pre-filing behavior relevant to whether post-filing damages should be enhanced. Therefore, because Radware alleges pre-filing conduct in addition to post-filing conduct, Se,agate is not controlling and Radware did not need to file a preliminary injunction in order to seek enhanced damages for willfulness.

Third, F5 argues that is entitled to summary judgment on willfulness because it has objectionably reasonable defenses. F5 Damages MSJ at 4-5. As discussed earlier, F5 may defend a willfulness claim by showing objectively reasonable defenses on infringement or invalidity. However,

[a]s the party moving for summary judgment [F5] must do more than persuade [the court] that its defenses were reasonable, Instead, [F5] must establish that there is no genuine dispute as to any material fact and that [the accused infringer] is entitled to judgment as a matter of law — in other words, that no reasonable fact-finder could find willful infringement.

HTC Corp. v. Tech. Props. Ltd., Case No. 08-0882, 2013 WL 5225043, at *8 (N.D.Cal. Sept. 17, 2013) (internal citations omitted) (citing Fed. R. Civ. P. 56(a)). As explained in this order, the court has granted summary judgment of infringement of certain claims of the ’319 Patent based on pre-hotfix product. The court also denied in its entirety F5’s Motion for Summary Judgment of Invalidity. F5 has not shown that no reasonable fact finder could find willful infringement. Because F5 has not met the high burden required to establish that no reasonable jury could find willful infringement, the Court DENIES F5’s motion for summary judgment on willfulness.

C. Damages for Lost Profits

Radware seeks damages in the form of lost profits of 12% of F5’s GTM and LTM sales.. Specifically, Radware’s expert-Mr. Malackowski opined that:

Radware and its asserted technology meet the Panduit test; therefore, 100 percent of the sales are appropriately considered for lost profits.

I have reduced that analysis by 88 percent to use what I believe tó be the best available evidence as a conservative measure of what apportionment would apply in a lost profits analysis.

Dkt. No. 204-31 (Malackowski Dep.) at 57:11-19; see also Dkt. 185-3 (Malackowski Rep.) at 72 (“it is reasonable to conclude that at least 12% of F5 customers which purchased BIG-IP [LTM or GTM] devices would have, alternatively, turned to Rad-ware in order to obtain this important functionality but for F5’s infringement.”). F5 moves to exclude Radware’s lost profits theory on the basis that Radware’s evidence showing that Radware would be entitled to 12% of its lost profits is unreliable. The court agrees that Radware’s 12% *1014figure is not supported and GRANTS the motion for summary judgment of no lost profits on LTM and GTM sales.27

1. Legal Standard

' To recover lost profits, a patentee “must prove a causal relation between the infringement and its loss’ of profits” by demonstrating a ’ reasonable probability that that “‘but for’ the infringement, it would have made the infringer’s sales.” BIC Leisure Prods., Inc. v. Windsurfing Int'l, Inc., 1 F.3d 1214, 1218 (Fed.Cir.1993). A lost profits award cannot be speculative. Id. The Panduit factors-are commonly, though not exclusively, used to show “but for” causation in a lost profits analysis by demonstrating: (1) demand for the patented product; (2) the absence of acceptable non-infringing substitutes; (3) manufacturing and marketing capability to exploit the demand; and (4) the amount of profit the patent owner would have made. See Standard Havens Prods., Inc. v. Gencor Indus. Inc., 953 F.2d 1360, 1372-73 (Fed.Cir.1991); Panduit Corp. v. Stahlin Bros. Fibre Works, 575 F.2d 1152, 1156 (6th Cir.1978).

2. Demand for the Patented Product Versus Demand for the

Patented Feature

Radware and F5 dispute whether the first Panduit factor requires showing demand for the patented product or demand for the patented feature. The Federal Circuit resolved this dispute in DePuy Spine, Inc. v. Medtronic Sofamor Danek, Inc., 567 F.3d 1314, 1330 (Fed.Cir.2009). The Federal Circuit explained that “[t]he first Panduit factor simply asks whether demand existed for the ‘patented product,’ i.e., a product that is ‘covered by the patent in suit’ or that ‘directly competes with the infringing device.’ ” 567 F.3d at 1330 (citing Rite-Hite Corp. v. Kelley Co., 56 F.3d 1538, 1548-49 (Fed.Cir.1995) (en banc)). The Federal Circuit rejected the argument that the first Panduit factor requires showing demand for the patented feature because that “argument unnecessarily conflates the .first and second Pan-duit factors,.. [the first] factor does not require any. allocation of consumer demand among the various limitations recited in a patent claim.” Id. Instead, whether demand exists for the patented feature is analyzed either under the second Panduit factor — the existence of non-infringing alternatives — -or when the patentee seeks to invoke the entire market value rule in the context of lost profits. Id. at 1330-31. The Federal Circuit explained this, in depth, distinguishing the primary cases on which F5 relies:

Ferguson dealt with lost profits under the “entire market value” rule, which permits a patentee to recover the entire value of an apparatus that contains both patented arid unpatented components, so long as the patented component is the basis for customer demand. Ferguson Beauregard/Logic Controls, Div. of Dover Res., Inc. v. Mega Sys., LLC, 350 F.3d 1327, 1346 (Fed.Cir.2003) (vacating and remanding lost profits award for entire value of a device containing a first component embodying a first patent, found infringed, as well as a second component' embodying a second patent, found not infringed, where profits could fairly be allocated to customer demand for second component); see Rite-Hite, 56 F.3d at 1549 (“[T]he entire market value rule' permits recovery of damages based on the value of a patentee’s entire apparatus containing several features when the patent-related feature is the ‘basis for customer demand.’ ”) (quoting State Indus., Inc. v. Mor-Flo Indus., *1015Inc., 883 F.2d 1573, 1580 (Fed.Cir.1989)).

DePuy, 567 F.3d at 1331.

Here, Radware argues that it is not relying on the entire market value rule in support of its lost profits analysis. That argument is accurate to the extent that Radware is not seeking 100% of its lost profits on 100% of F5 sales. However,. Radware’s expert testified that “Radware and its asserted technology meet the Pan-duit test; therefore, 100 percent of the sales aré appropriately considered for lost profits,” suggesting the entire market value rule might apply. Dkt. No. 204-31 (Ma-lackowski Dep.) at 57:11-19. To the extent that Radware relies on the entire market value rule to recover lost profits, Radware must show what portion of sales were driven by demand for the patented feature, consistent with application of the entire market value rule in general. See DePuy, 567 F.3d at 1331; see also Good Tech. Corp. v. MobileIron, 2015 WL 3882608 at *4 (N.D.Cal. June 23, 2015) (“The EMVR requires that where damages are based on sales of a multi-component product that includes both infringing and non-accused components, in the absence of an apportionment, á patentee must prove that the patented features are the primary' driver of demand for the entire product.”). Rad-ware has not- met the burden of. showing what portion of sales were driven by demand for the patented feature. Radware’s expert explained in his report that 12% of customers identified the patented functionality as important. Malackowski Rep. at 72. A survey showing that 12% of people found a feature important is not, standing alone, proof that the feature drove those individuals’purchases. .

If Radware is simply using the Panduit factors to show the number of sales Rad-ware would have made but for F5’s infringement, Radware may. either show the absence of non-infringing alternatives under the second. Panduit factor, or show what portion of customers would have rejected available non-infringing alternatives and purchased Radware’s products. See Northern District of California Model Patent Jury Instruction; 5.3 (“Lost, Profits— Factors to Consider”) (Second factor: “that there were no acceptable non-infringing substitutes for the [product] [method] for which [patent holder] seeks lost profits, or, if there were, the number of sales made by [alleged infringer] that [patent holder] would have made despite the availability of any acceptable non-infringing substitutes.”) (emphasis added). Radware is attempting to do the latter: to show the number of sales made by F5 that Radware would have made despite the availability of acceptable non-infringing substitutes.

■ The court next examines whether Rad-ware’s evidence purporting to show that Radware would have recovered 12% of the sales of GTM and LTM is sufficiently reliable to justify an award.

3. Radware’s Evidence Does Not Support a 12% Lost Profit Award

As just explained, Radware must provide reliable evidence that but for F5’s sales of the BIG-IP products, Radware would have made sales’ of its own device. Radware’s expert opines that Radware would have recovered 12% of F5’s sales, once F5 stopped infringing Radware’s patents. ' Dkt. 185-3 (Malackowski Expert Rpt.) at 71-73. The court finds Radware’s evidence supporting a 12% lost profit award legally insufficient.

Radware attempts to justify its lost profits calculation based on a 2011 survey conducted by TechValidate.28 The survey *1016targeted F5 customers that used an Oracle solution and had 261 respondents. Id. at 71. The survey listed seventeen features of F5’s BIG-IP product, including link load balancing, compression, and advanced routing, and asked customers to select any and all features that they considered important to performance. Id. at 71-72. 12% of respondents selected link load balancing as an “important” feature. Based oh these results, Malackowski concludes that 12% of F5 customers who purchased BIG-IP-GTM devices would have alternatively purchased Radware products but for F5’s infringement. Id. at 72. However, the survey demonstrates that multiple features, including features unrelated to the Asserted Patents, drive the demand for the accused products. See id. For example, 72% of participants selected “local load balancing methods” and 48% of participants selected “SSL offload” as functionalities important to performance. Id. Simply put, the survey does not support the conclusion that had F5 exited the market, 12% of F5 customers would have túrned to Radware. Here, the TechValidate survey does not in any way suggest the patented feature was the primary reason that customers purchased the infringing product. In other words, the survey is not evidence that but - for the infringing sales, some consumers would have instead purchased Radware’s products for the reason that they offered the infringing feature.

Moreover, Malackowski’s analysis does not attempt to account for the. different features between, the Radware products and F5 accused .products. The TechVali-date survey listed some functionalities that are proprietary to F5, such as iRules and iControl. See Malackowski Rep. at 72; 184-4 (Brewer Decl.) ¶ 18. But Malackowski’s report offers no evidence addressing the significance of these proprietary functions to F5 customers. Interestingly, the survey shows that 39% of participants considered the iRules function important to performance. See Malackowski Rep. at 72. F5 points out that under Malackowski’s reasoning, 39% of survey participants would not have bought a Radware' product because it did not contain the iRules function, suggesting that the Radware and F5 products are not interchangeable. F5 Damages MSJ at 17.

Because Malackowski cannot rely on the TechValidate survey to show that Radware would have made 12% of F5’s LTM and GTM sales “but for” F5’s infringement, the court. GRANTS F5’s motion for summary judgment as to Radware’s entitlement to seek .lost profits.

VIL CONCLUSION

For the reasons explained above, the court orders as'follows: '

• Radware’s Motion to Strike is:

• GRANTED as to F5’s Section 101 invalidity defense and F5’s Motion for-Judgment on the Pleadings;

• DENIED as to those portions of F5’s Motion for Summary Judgment that rely upon GSLB;

• GRANTED as to those portions of . F5’s Motion for Summary Judgment that rely upon BGP;. and

• DENIED- as to the declaration of Dr. Peter Alexander.

• F5’s Motion1 for Judgment on the Pleadings is DENIED.

• F5’s Motion to Amend its. Invalidity Contentions is DENIED.

• Radware’s-Motion for Sanctions under Rule 37 is DENIED subject to - the conditions specified'.

• Radware’s Motion for Summary Judgment' regarding F5’s invalidity affirmative defense is:

• DENIED as to the allegedly abandoned prior art references;

*1017• GRANTED as; to the testbed systems as prior art;

• DENIED as to the Maki-Kullas patent; and

• DENIED as to the Cisco DD white paper references.

• F5’s Motion for "Summary Judgment of Invalidity is: .

• DENIED that,, GSLB anticipates the Asserted Patents;

• DENIED that BGP anticipates the Asserted Patents; and

• DENIED that the Asserted Patents are obvious.

• On the infringement motions:

• Per the parties’ agreement, F5’s Motion for Summary Judgment of no direct infringement of the asserted method claims29 is GRANTED (pre- and post-hotfix).

• Radware’s Motion for Summary Judgment of Infringement is GRANTED as to claim 24 of the ’319 Patent (pre-hotfix).

• F5’s Motion for Summary Judgment of Non-Infringement is GRANTED as to ’319 Patent claim 10 (pre- and post-hotfix direct infringement).

• Both parties’ motions are DENIED as to the remaining clairas/products.

• F5’s Motion for Summary Judgment on damages is GRANTED as to pre-complaint damages and lost profits on the LTM and GTM sales and DENIED as to willfulness on pre-hotfix products.

• F5 has 30 days from the date of this order to make its updated, post-hotfix source code available for inspection to Radware and provide Mr. Thornewell or a mutually agreed witness to testify about it. At that point, F5 may renew its Motion for Summary Judgment of Non-Infringement as to post-hotfix products. . ;

• The court has filed an hnredacted copy of this order under seal. If a party believes that" the redacted portion discloses confidential information, it must file a version of the order with proposed redactions and provide a declaration setting forth the bases for asserting confidentiality. The declaration and proposed redactions may be filed under seal. The court will evaluate any such confidentiality contention and make a decision whether to. approve the proposed redaction or to remove it, thus rendering the underlying content public. Any proposed redactions and declarations in support must be filed by October 26, 2015. Any: responses to the proposed redactions must be filed by November' 3, 2015. Thereafter, a "copy of this order with the approved redactions will be made public.

IT IS SO ORDERED.

. DNS is an acronym for "Domain Name System.” See, e.g., Dkt. No. 179-26 ¶ 51.

. To be clear, F5 would not be able to assert another embodiment of GSLB as anticipating prior art at trial, nor would "it be áble to simply argue that GSLB, unmoored from its implementation in Cisco DD, anticipates.

. Dr. Rubin’s expert report regarding BGP in full states that:

37. A number of the systems and references that Dr. Alexander relies on relate to - Border Gateway Protocol (BGP), or require BGP; a technology unrelated to the Rad-ware Patents. The link load balancing systems and methods disclosed in the Radware Patents do not relate to BGP. The solutions taught by the Radware Patents for inbound and outbound link load balancing were devised as an improved solution to multihom-ing [rather] than BGP. During prosecution, the patent applicants - made clear that the multihoming solution described by the Rad-ware Patents did not relate to BGP:

Additional, other prior art submitted in an IDS currently with this preliminary amendment describes a multi-homed environment, which is based on BGP. The BGP mechanism fails to suggest the feature of the applicant’s claimed invention of resolving a DNS resolution query in the context where the device is connected to the Internet through a plurality of routes and the plurality of routes are assigned with respective IP addresses.

38; Thus, from the outset, the patent applicants distinguished their load balancing solution from BGP. Dr. Alexander’s reliance on BGP is misplaced because Radware’s solution solves a multihoming problem in a different way,

Dkt. No, 213-2 at ECF p.4.

. In its reply, Radware also moved in limine to exclude the use of the test-bed systems as demonstrative evidence that the prior art 3-DNS and Cisco DD systems could perform ISP load balancing. Dkt. No. 209 at 12. However, the court disagrees with Radware that the issue has been fully presented in the context of this briefing, and declines to reach the issue at this juncture.

.The '319 Patent was filed on June 2, 2003, but it enjoys the filing -date of December 20, 1999, because the '319 patent is a division of application No. 09/467,763 (now Patent No. 6,665,702). Dkt. 24 at Ex.' B. The '374 Patent was filed on August 3, 2012, but also enjoys the filing date of December 20, 1999, because it is a continuation of the '319 Patent applica- ., tion filed on June 2, 2003, which in turn is a division of application N°- 09/467,763, filed on December.20, 1999. See id. at Ex. C.

. F5’s expert, Dr. Alexander, conceded that the Maki-Kullas patent is Section 102(e) art and thus not entitled to a priority date earlier than its October 5, 1999 filing date. Dkt. No. 189-7 at 163; Dkt. No. 186-10 at' 160:15-161:6.

. Dr. Rubin’s full conception opinion states:

[redacted]

., Radware’s attempt to distinguish Invitrogen fails. See Dkt. No. 209 at 5. The court’s discrediting of the proffered expert testimony in that case was not due to the fact that the inventor's notebook entries contradicted his deposition testimony, as Radware claims. Id. Rather, the court found that "Clontech nowhere provides the court with expert testimony that properly explains the technical notebook entries- advanced in support of its conception arguments.” Invitrogen, 429 F.3d at 1068. The expert’s declaration in Invitro-gen provided the court "no substantive factual guidance,” as the expert stated .only that “It is clear from [the inventor's] testimony and his lab records that he conceived of the invention in 1984.” Id. Like Dr. Rubin, the expert in Invitrogen cited various documents (in that case, notebook entries), and concluded that "[t]hese representative entries demonstrate Goff’s conception, diligence and reduction to practice.” Id. The court found that “such wholly conclusory assertions on a legal issue cannot carry [a party’s] burden on summary judgment,” and the court so finds in this case as -well.

. Radware also offers the expert testimony of Radware’s technical expert, Dr. Rubin. Dkt. No. 186-4 at 19. However, this testimony does not constitute evidence of reduction to practice for the same reason Dr. Rubin’s conception opinion failed to do so: Dr. Rubin merely cites Radware’s other evidence, without adding any substantive analysis or explanation, before stating his conclusory opinion that after [redacted) Dkt. No. 186-24, ¶ 40.

. The parties agree that Radware constructively reduced the invention to practice on December 20, 1999, the date on which it filed the '702 patent application, See Dkt. Nos, 189 at 19¡ 196-4 at 7.

. In its administrative motion seeking leave to file portions of its reply brief under seal, Radware identifies Cherkassy as ‘‘an engineer at Radware who offered testimony regarding the conception and reduction to practice of the" claimed inventions.” Dkt. No, 209-1 at 2.

. As the court concludes that Rádware has failed to show that it reduced the inventions disclosed in the ’319 and ’374 patents to practice by October 4, 1999, the court need not reach the parties’ arguments regarding diligence.

. F5 also moves for summary judgment that the asserted outbound claims in both patents are invalid as anticipated by BGP. Dirt. No. 183 at 20-21. However, because the court granted Radware’s motion to strike those-portions of F5’s motion for. summary judgment that rely on BGP, F5’s motion for summary judgment that BGP anticipates is DENIED.

. The court previously construed "Proximity/proximities” to mean “a measurement or measurements based on hops, latency, TTL, or a combination thereof.” Dkt. No. 122 at 6. The court also construed /'Costing information” and “cost” as “plain and ordinary meaning [but] not limited to monetary price.” Id. at 25-26. The court observed that "cost” “can represent an array of non-monetary resources. Something can ‘cost’ bandwidth or be expensive in terms of time.” Id. at 26.

. The parties dispute the relevance of the presence of a networking switch in the testbed system, as well as configuration files written by Dr. Alexander in 2014.

. In fact, Dr.' Rubin testified that GSLB cannot even perform outbound server load balancing, let alone link load balancing. Dkt. 195-1 ¶ 6.

. Because the court finds that F5 has not met its burden of showing obviousness by clear and convincing evidence, the court need not reach objective indicia of nonobviousness.

. The experts disagree on whether the remaining functionality is infringing, as dis*1000cussed below.

. The asserted method claims are '319 Patent . claims 13-23 and 26-28; ’374 Patent claims 1-4 and 6-7.

. The court emphasizes that the products "may infringe” because the parties have not moved for summary judgment on all claims. Furthermore, as discussed below, the court grants summary judgment of non-infringement only as to certain claims.

. Claims 26-28 of the '319 Patent are method claims, and, as noted above, the court grants F5’s motion for summary judgment of no direct infringement of method claims.

. As noted earlier, the '319 Patent and the '702 Patent share the same specification.

. According to Radware’s expert,, the "Observed” and "Predictive”, methods also rely on the number of connections, at a particular node. See Dkt, No. 179-26 ¶ 128. Accordingly, these methods also only measure load on a portion of a route.

. At the hearing, F5 suggested that claim 12 of the '319 Patent, which recites "the device of claim 1, wherein the plurality of routes each include a respective Internet Service Provider ("ISP”), is not infringed for similar reasons.- However, this -argument was- not made in F5’s briefing. Accordingly, the court does not consider it.

. The court GRANTS Radware's Motion for Relief to File Objection to Reply Evidence. Dkt. No. 226. However, the court OVERRULES Radware’s objection to F5’s reply evidence because the evidence submitted with F5’s reply, Dkt. Nos. 215-7 and 215-8, merely provides further context for evaluating the evidence presented in F5’s opening brief. See. e.g., Dkt. No, 187.

. This order does not address lost profits related to Link Controller sales.

. Although Mr. Malackowski testified that he relied on “the totality of the information in [his] report,” the only evidence cited in support of the 12% figure is the TechValidate survey. Compare Malackowski Dep. at 52:19-20 to Malackowski Rep. at § 10.4.1.2.

. ’319 Patent claims 13-23 and 26-28; ’374 Patent claims 1-4 and 6-7.