Radware, Ltd. v. F5 Networks, Inc.

Full title: RADWARE, LTD., et al., Plaintiffs, v. F5 NETWORKS, INC., Defendant.

Court: UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA SAN JOSE DIVISION

Date published: Feb 13, 2016

Case No. 13-cv-02024-RMW (N.D. Cal. Feb. 13, 2016)

Opinion

Case No. 13-cv-02024-RMW

02-13-2016

RADWARE, LTD., et al., Plaintiffs, v. F5 NETWORKS, INC., Defendant.

Ronald M. Whyte United States District Judge

ORDER REGARDING MOTIONS IN LIMINE

Re: Dkt. Nos. 355, 362, 364, 369-373, 377, 380, 380-1, 411, 415

The court held pretrial conferences in this matter on February 4, 2016 and February 11, 2016. In advance of those conferences, Radware submitted 16 motions in limine as well as a motion to exclude testimony from F5's expert witnesses, and F5 submitted 8 motions in limine.

I. F5'S MOTION IN LIMINE NO. 1: TO EXCLUDE EVIDENCE OF THE TECHVALIDATE SURVEY

DENIED. F5 seeks to exclude a survey from TechValidate that Radware's damages expert James Malackowski relies upon in support of his lost profits analysis. Dkt. No. 355. The survey targeted F5 customers that used an Oracle solution and had 261 respondents. Dkt. No. 274-17 at 71. The survey listed seventeen features of F5's BIG-IP product, including link load balancing, compression, and advanced routing, among others, and asked customers to select any and all features that they considered important to performance. Id. at 71-72. Twelve percent of respondents selected link load balancing as an "important" feature.

Radware explains that it intends to use the survey in support of Panduit factors 2 (the absence of acceptable noninfringing substitutes) and 4 (the amount of profit).¹ Dkt. No. 446 at 3-7. Specifically, Radware argues that the survey allows Radware to reconstruct the market for customers who find Radware's patented technology "important" as that market would have developed in the absence of infringement:

1 Panduit Corp. v. Stahlin Bros. Fibre Works, Inc., 575 F.2d 1152 (6th Cir. 1978)

In the Panduit reconstructed market, Radware would be the only player offering a link load balancing product, and, under the classic Panduit approach, would capture 100% of the sales in that reconstructed link load balancing market. In other words, Radware contends that it is entitled to 100% of the 12% of F5 sales that constitute the reconstructed LLB market.

Id. at 4. While F5 notes that Radware's expert only cites the survey in support of Panduit factor 4, F5 agrees with Radware that the survey's conclusion actually belongs under Panduit factor 2. Dkt. No. 444 at 7.

F5 argues that Radware and Mr. Malackowski fail to provide details of the survey, including how it was conducted, who the participants were, or which F5 software modules they licensed. See Dkt. No. 355 at 1. F5 also argues that because the survey merely asks participants to select all features of F5's products that the participants found "important," the survey does not reveal which features were most important or drove users' purchases. See id. Accordingly, F5 argues, the survey does not suggest that had F5 exited the market, customers would have gone to Radware.

Radware responds that, in fact, F5 possesses the additional details regarding the participants of the survey that Radware claims are absent, Dkt. No. 367 at 3, and F5 regularly cites TechValidate case studies in F5's own marketing materials, id. at 2-4. On January 28, 2016, F5 filed a reply brief in support of its motion. Dkt. No. 395.² In its reply, F5 notes that the fact that F5 has cited other TechValidate surveys in the past does not render this specific TechValidate survey admissible. F5 concedes that F5 possesses additional details on the survey participants. Id. at 2-3. This undercuts F5's argument that the lack of information about respondents renders the survey unreliable. Nevertheless, F5 argues, the additional details show that the survey does not indicate demand for link load balancing among users of F5's GTM and LTM modules. F5 points out:

2 Only the following day did F5 seek leave to file its reply. Dkt. No. 411. While the court's scheduling order, Dkt. No. 279, did not allow replies in support of motions in limine, because Radware filed a surreply that addressed the points in F5's reply, the court grants F5's motion for leave to file a reply brief, Dkt. No. 411, and Radware's motion for leave to file a surreply, Dkt. No. 415.

(1) all participants who selected "link load balancing" also identified other functionalities before it, including at least five participants who identify iRules, a proprietary F5 function; (2) there is no indication of whether "link load balancing" was defined or whether it was interpreted as limited to the patented functionalities; (3) some participants who identified link load balancing did not even indicate that they had licensed GTM or LTM; (4) some identified themselves as LTM licensees, which has link load balancing functionality that is not accused of infringement; (5) some did not identify what products they had licensed; and (6) some identified themselves as Link Controller licensees, where link load balancing is the primary functionality.

Id. at 3 (citations omitted). F5's reply also attacks the survey's reliability in the "but for" test for causation of lost profits and largely repeats arguments made in the parties' summary judgment briefing on lost profits.

Based on the parties' submissions, the court finds that F5's challenges to the TechValidate survey go to its weight and not to its admissibility. To be sure, the survey provides extraordinarily weak support for Radware's argument that 12% of F5's LTM and GTM customers would have bought from Radware had infringing versions of LTM and GTM been removed from the market. The fact that a customer found a feature "important" does not necessarily mean that the feature drove that customer's purchase. However, a reasonable juror could rely on the survey results, in combination with the underlying data and other independent evidence, to conclude that some nonzero number of LTM or GTM customers found link load balancing important enough to drive their purchases. The court need not decide at this stage whether any proffered evidence of lost profits other than the survey is admissible or whether the evidence ultimately submitted to the jury is sufficient to meet Radware's burden of proof. II. F5'S MOTION IN LIMINE NO. 2: TO PRECLUDE DR. RUBIN FROM OPINING AS TO CONCEPTION AND REDUCTION TO PRACTICE, THE DOCTRINE OF EQUIVALENTS, AND INDIRECT INFRINGEMENT

GRANTED IN PART. F5 seeks to preclude Dr. Rubin from opining on certain topics at trial.

A. Conception and Reduction to Practice

F5 seeks to preclude Dr. Rubin from opining on conception and reduction to practice. In his report, Dr. Rubin states that his opinions are based on documents, deposition testimony, and Radware's infringement contentions. F5 argues that he should not be permitted to opine on these topics under Federal Rule of Evidence 702 and because of the analytical gap between the evidence listed by Dr. Rubin and his ultimate conclusions. Dkt. No. 362 at 4 (citing Magnetar Techs. Corp. v. Six Flags Theme Parks Inc., No. CV 07-127-LPS-MPT, 2014 WL 529983, at *12 (D. Del. Feb. 7, 2014)). F5 further argues that Dr. Rubin's report fails to provide a complete statement of the opinions on these topics as required by Federal Rule of Civil Procedure 26(a)(2)(B). Dkt. No. 362 at 4-5.

F5's motion is granted. The opinions in Dr. Rubin's report consist solely of citations to evidence and his conclusions on conception, reduction to practice, and diligence. Dr. Rubin did not perform any technical analysis that would assist the jury. Although contentions as to conception and reduction to practice do not require claim charts under the patent local rules, Dr. Rubin was required to offer something to close the analytical gap between the evidence and his opinions.

B. Doctrine of Equivalents and Indirect Infringement

F5 seeks to prevent Dr. Rubin from opining on infringement under the doctrine of equivalents and indirect infringement. At the pretrial conference, the parties agreed that the only remaining infringement issue would be decided by the court. Therefore, these motions are moot.

III. F5'S MOTION IN LIMINE NO. 3: TO EXCLUDE CERTAIN OPINION TESTIMONY OF PLAINTIFF'S TECHNICAL EXPERT DR. RUBIN

GRANTED IN PART AND DENIED IN PART. F5 argues under Federal Rule of Evidence 702 and the principles of Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993), that Radware's technical expert Dr. Rubin should be precluded from offering opinion testimony regarding (1) the purported "importance" of link load balancing and (2) various nontechnical objective indicia of nonobviousness, including commercial success, long-felt need, industry praise, failure of others, and copying. Dkt. No. 364.

A. Legal Standard

Federal Rule of Evidence 702 governs the admissibility of expert testimony. An expert must be qualified by virtue of his or her "knowledge, skill, experience, training, or education." FRE 702. Expert testimony must also be helpful to the trier of fact in understanding the evidence or determining a fact in issue. Id. The court must be convinced that the expert testimony is reliable. Id. Reliable testimony must be (1) "based upon sufficient facts or data," (2) "the product of reliable principles and methods," and (3) a sound application of the principles to the facts. Id. "Facts or data" may include other experts' reliable opinions or hypothetical facts that are supported by the evidence. FRE 702, Adv. Committee Note (2000); see FRE 703. The party offering expert testimony must demonstrate by a preponderance of the evidence that its expert's opinions are reliable. See In re Paoli R.R. Yard PCB Litig., 35 F.3d 717, 743-44 (3d Cir. 1994). The Supreme Court has confirmed that "the Rules of Evidence—especially Rule 702—do assign to the trial judge the task of ensuring that an expert's testimony both rests on a reliable foundation and is relevant to the task at hand." Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 597 (1993).

B. Analysis

F5 concedes that Dr. Rubin, an electrical engineer, is qualified to testify regarding the technical aspects of the server and link load balancing technologies that he disclosed in this case. F5 argues, however, that the opinions in Rubin's expert reports stray outside of his expertise and usurp the role of the jury. See id. at 1-2. Radware responds that Dr. Rubin has relevant industry experience, that all of his opinions are technical in nature, and that his opinions are based on reliable methodologies. Dkt. No. 407 at 1. The court addresses the parties' arguments under each area in which F5 seeks to limit Dr. Rubin's testimony.

1. Importance of Link Load Balancing

F5 first argues that Dr. Rubin is not qualified to opine on the importance of link load balancing because he has not surveyed the market and because, when asked at a deposition, he could not identify a single company that uses link load balancing based on his personal knowledge. Dkt. No. 364 at 4 (citing deposition testimony). Accordingly, F5 moves to exclude the testimony at paragraphs 64-66 of Rubin's initial expert report (Dkt. No. 364-2). Radware responds that Rubin's opinions on the importance of link load balancing are technical in nature and that in support of his opinions, he actually relies on an article by an F5 "subject matter expert" that identifies reliability as a driver for the adoption of link load balancing. Dkt. No. 407 at 7-9.

The court finds that because Dr. Rubin is unaware of any companies that actually use link load balancing, any opinion from Rubin regarding the importance of link load balancing to particular companies would be speculative and unreliable. However, he could explain from his experience how link load balancing works and how it is important in today's business world.

2. Commercial Success and Industry Praise

F5 moves to exclude Dr. Rubin's opinions on commercial success and industry praise as conclusory. Rubin's opinions on these topics are provided in two paragraphs of his rebuttal report, reproduced below:

E. Industry Praise For Radware's Patented LLB Technology

58. Radware has received numerous awards for LinkProof. For example, in 2003, LinkProof was awarded the Editor's Choice award by Network Computing Magazine.[citation]

F. Commercial Success of Radware's Patented LLB Technology

59. In my Initial Report, I identify the Radware Products that practice the Radware Patents, as well as the F5 products that infringe the Radware Patents. As I am informed that the Radware Products that practice that [sic] Radware Patents, and the F5 products that infringe the Radware Patents, have been commercially successful,[citation to Malackowski Report] the success of both Radware's Products and F5's infringing products further refutes Dr. Alexander's opinion that the LLB solutions covered by the Radware Patents were obvious.

Dkt. No. 186-25 ¶¶ 58-59.

F5 argues that Dr. Rubin's opinion on commercial success is "little more than a citation to Radware's damages report" and that Rubin is not qualified to testify about why a given product was or was not commercially successful. Dkt. No. 364 at 7-8. Moreover, F5 argues that Rubin did not conduct an analysis of commercial success at all, let alone one based upon a reliable methodology. Id. at 8. Radware responds that Rubin's testimony on commercial success is "purely technical" and involves linking Radware's patented product to the claimed invention. Dkt. No. 407 at 20. To the extent that Radware's is referring to the sentence in which Rubin explains how his initial report identifies the Radware products that practice the Radware patents as "technical" testimony, this court agrees with Radware. It is well within Dr. Rubin's expertise to compare the asserted claims to actual products. The remainder of Dr. Rubin's opinion on commercial success, however, is entirely conclusory. While reliance on Radware's damages expert Mr. Malackowski is not improper in and of itself, a citation to the Malackowski report, without more, does not assist the jury in evaluating F5's obviousness arguments.

Dr. Rubin's opinion on industry praise fails for similar reasons. Rubin offers no independent basis for his opinion and instead simply cites an article indicating that the editors of a magazine praised Radware's LinkProof product. Even if Radware is correct that Dr. Rubin established elsewhere that LinkProof is coextensive with the patents in suit, Rubin's opinion on industry praise offers no analysis whatsoever. Nothing in this paragraph is intended to preclude Radware from offering evidence that it received numerous awards for LinkProof. The court's ruling is merely intended to limit Dr. Rubin's testimony to the contents of his expert report.

Accordingly, the court grants F5's motion to exclude Dr. Rubin's opinions on the issues of commercial success and industry praise. Dr. Rubin may testify regarding technical subjects such as whether a product embodies a claimed invention or whether a product is coextensive with a claimed invention. To the extent that Radware wishes to present evidence of commercial success and industry praise, however, it will need to come from some other source.

3. Long Felt Need and Failure of Others

F5 argues that the paragraphs of Dr. Rubin's rebuttal report dedicated to long felt need comprise a handful of citations to publications discussing the release of Radware's LinkProof product combined with a conclusion—without any analysis—that there was a long felt need in the market for the invention disclosed by the Radware patents. See Dkt. No. 364 at 9 (citing Dkt. No. 186-25 ¶¶ 44-49). F5 contends that the cited evidence is insufficient for Rubin to present an opinion on long felt need. In response, Radware argues that when the paragraphs of Dr. Rubin's report under the heading of long felt need are examined in context with the sections of his rebuttal and initial reports describing the failure of Border Gateway Protocol ("BGP"), one must conclude that he has presented a sufficient basis to opine on long felt need. See Dkt. No. 407 at 17-18. Accordingly, this order addresses Rubin's opinions on long felt need and failure of others together.

Under the heading of "Failure of Others to Devise a Multi-Homing Solution," Dr. Rubin opines that before Radware's invention of LinkProof, the only known mechanism for managing a multi-homed network was BGP. Dkt. No. 186-25 ¶ 51. Rubin argues, however, that "BGP is not a viable solution for multihoming for reasons which includes cost, complexity, and inefficiency, among others." Id. Rubin asserts that "[s]everal white papers published by F5 and cited in [his] Initial Report confirm [his] opinion that BGP is not a viable alternative or solution to the problem that LLB solves." Id.

F5 criticizes Rubin's opinions regarding the failure of others as unsupported by sufficient data or reliable methodologies. Dkt. No. 364 at 10-11 (moving to strike paragraphs 50-52 of Rubin's rebuttal report). F5 argues:

Dr. Rubin does not explain why he believes that BGP represents efforts to solve the problem of the claimed inventions—in his words, "to use a single device solution to solve in a reasonably simple manner the distribution of load among ISP links"—rather than some other problem relating to network configuration and routing.

Id. at 11 (citing Dkt. No. 363-6 (Rubin Dep. Tr.) at 80:3-5). In response, Radware points to the analysis of two F5 whitepapers in Dr. Rubin's initial report. Dkt. No. 407 at 14-15 (citing Dkt. No. 179-26 (Rubin Initial Rep.) ¶¶ 67-75). According to Rubin, a key message of the whitepapers is that BGP is not well suited to provide multi-homing. Dkt. No. 407 at 15 (citing Dkt. No. 407-6 at 1). The court finds that Radware has presented a sufficient basis for Dr. Rubin to express his expert opinion that BGP was a failed attempt to manage links in a multi-homed environment. Accordingly, F5's motion to exclude Dr. Rubin's opinions on failure of others are denied.

The court thus returns to the issue of whether to allow Dr. Rubin's testimony regarding long-felt need. Dr. Rubin's analysis of the articles cited in paragraphs 44-49 of his rebuttal report describes the advantages of Radware's LinkProof product and indicates that LinkProof was the first device of its kind. Dr. Rubin's report provides relatively little analysis on whether these articles—or other evidence—show that people actually needed link load balancing before Radware's patented products entered the market. Nevertheless, at least some of Dr. Rubin's citations suggest that people used multi-homed environments and that they were looking to address problems with those environments. For example, Rubin cites an article from the South China Morning Post indicating that "[u]sing multiple ISPs adds redundancy abilities but necessitates complex configuration and routing protocols as well as close coordination between the contracted ISPs." Dkt. No. 186-25 ¶ 46 (citing Dkt. No. 407-8). Rubin also cites a different article from NetworkMagazine.com that, according to Rubin, "describes the complexities and pitfalls of trying to use BGP for multihoming." Id. ¶ 45 (citing Dkt. No. 407-10). The court finds that Radware has provided a sufficient foundation for Dr. Rubin to provide testimony regarding long felt need. Accordingly, F5's motion to exclude Rubin's opinions on long felt need is denied.

4. Copying

Dr. Rubin's rebuttal report cites several internal F5 documents that mention Radware and/or LinkProof and suggest—sometimes quite explicitly—that F5 was trying to develop a product that competed with LinkProof. Dkt. No. 186-25 ¶ 53. For example, Rubin asserts that in one document, an F5 employee wrote that "F5 needs to develop a solution that . . . adds critical ISP LB functionality to BIG-IP and 3-DNS to better address customer needs" and that "F5 is trying to displace a more established product in Radware's LinkProof." Id. (citing Jason Needham Depo. Tr., Exhibit 2).

F5 argues that Dr. Rubin has no foundation to opine on what F5 employees may or may not have relied upon when developing accused products. Dkt. No. 364 at 12. Further, F5 argues, Rubin's opinion on copying will not assist the jury. Id. Radware responds by arguing that Dr. Rubin's testimony is helpful in assisting the trier of fact to understand what LinkProof functionality F5 was looking to replicate, and how F5 replicated that functionality in creating its infringing product. Dkt. No. 407 at 13.

Both parties cite Rambus Inc. v. Hynix Semiconductor, Inc., 254 F.R.D. 597, 609 (N.D. Cal. 2008) in support of their positions. In Rambus, this court held that a technical expert could testify "as to what Rambus's disclosures revealed and whether any [defendant's] products or documents . . . contain substantially similar features." Id. However, this court found that the expert "lack[ed] a sufficient foundation for an opinion that any Manufacturer 'copied' Rambus's disclosure. Further, his conclusory opinion that some Rambus feature was 'copied' would not assist the jury 'to understand the evidence or to determine a fact in issue.'" Id.

The court reaches a similar conclusion in the instant case. Dr. Rubin may testify as to whether Radware's and F5's products have similar features, and Radware may argue that the evidence shows that F5 copied Radware's products. An ultimate opinion from Dr. Rubin that F5 copied Radware's products, however, will not assist the jury to understand the evidence or to determine a fact in issue.

IV. F5'S MOTION IN LIMINE NO. 4: TO PRECLUDE EVIDENCE AND ARGUMENT THAT THE A10 SETTLEMENT AFFECTED A10'S STOCK PRICE

GRANTED. F5 moves pursuant to Federal Rules of Evidence 702, 703, and 403 to preclude Radware's damages expert Mr. Malackowski or any other witness from testifying that Radware's settlement with A10 Networks caused A10's stock price to decline by over half of its value. Dkt. No. 369. Because the settlement agreement in question is confidential, large sections of the parties' briefing on the instant motion remain under seal. For purposes of the instant motion, however, it is sufficient to say that both sides' experts want to use the terms of the agreement in support of their damages arguments.

F5 notes that the terms of the settlement are not public today and were not public in September 2014 through October 2014 when A10's stock price dropped. Dkt. No. 369 at 1. Moreover, the agreement was only provided to Radware by A10 on October 28, 2015, after the stock price had fallen. Id. Accordingly, F5 argues that the settlement agreement could not have caused the drop in stock price. The court agrees that Radware has no basis to conclude that the settlement agreement hurt A10's stock price, and even if there were some basis, the risk of confusing the jury would far outweigh the probative value.

Radware agrees that testimony, evidence, or argument that the Radware-A10 settlement caused a drop in A10's stock price should be excluded but insists that F5's damages expert Laura Stamm should also be precluded from offering any testimony regarding the settlement. Dkt. No. 403 at 1. Radware did not bring a motion to exclude Ms. Stamm's testimony regarding the agreement. Nevertheless, the court agrees that Ms. Stamm's potential testimony, as outlined on page 1 of Radware's sealed response to F5's motion, Dkt. No. 401-6, is speculative and is likely to confuse the jury. Accordingly, Ms. Stamm will not be allowed to testify regarding the A10 settlement in support of her damages arguments. V. F5'S MOTIONS IN LIMINE NO. 5 AND 6 AND RADWARE'S MOTIONS IN LIMINE NOS. 9, 11, AND 14: USE OF PATENTS AND PATENT APPLICATIONS ON WILLFULNESS

A. F5's Motions No. 5 and 6

DENIED. In its fifth and sixth motions in limine, F5 seeks to exclude certain evidence in support of Radware's claim of willfulness, particularly evidence relating the '702 patent, evidence relating to the '319 application, and evidence of F5's alleged copying of Radware's LinkProof product to create the Link Controller product.

There are two remaining patents asserted in this case: the U.S. Patent No. 8,266,319 and U.S. Patent No. 8,484,374 patent. The asserted patents are descendants from the application that issued as U.S. Patent No. 6,665,702. Radware previously asserted the '702 patent in this case; the court granted summary judgment of noninfringement. According to Radware, evidence of F5's knowledge of the '702 patent and application that resulted in the '319 patent is relevant to the totality of the circumstances allegedly suggesting willfulness. Radware seeks to introduce evidence that F5 cited the '702 patent and the '319 application in the file histories of several of F5's own patent applications, including application No. 12/259/142 (the "Masters application"). Radware also seeks to introduce evidence that F5 copied Radware's LinkProof product.

This court determined that a material dispute of fact remained as to whether F5 knew about the '319 Patent prior to the filing of the complaint. This finding was based in part on evidence of F5's citation to the '702 patent and application for the '319 patent in prosecution of F5's own patents. This court also found that a jury could find evidence of F5's pre-filing behavior—including evidence of a "desire to enter the field of link load balancing"—relevant to whether post-filing damages should be enhanced. Therefore, Radware will be permitted to use F5's citation to the '702 patent and the application for the '319 patent in F5's Masters application, as well as evidence of F5's desire to enter the field of link load balancing, to show the totality of the circumstances allegedly suggesting willfulness.

B. Radware's Motions No. 9, 11, and 14

GRANTED IN PART. Radware's ninth, eleventh and fourteenth motions are related. Radware moves to exclude (1) any evidence of inequitable conduct as to the '702 patent and U.S. Patent No. 6,249,801, (2) any evidence regarding the reexamination of U.S. Patent No. 6,718,359, and (3) evidence that Radware delayed in bringing suit on the '702 patent. The applications for Radware's asserted patents were continuations-in-part of the application that lead to the '801 patent. The '359 patent issued from a separate continuation of the application that lead to the '801 patent. The '359 patent was later invalidated after reexamination. F5 would prefer that the court exclude all evidence related to these patents. In the alternative, F5 argues that to the extent Radware is permitted to rely on such patents to show willfulness, F5 should be able to present countervailing evidence to show, for example, that such patents are invalid or unenforceable or that F5 does not infringe.

At the pretrial conference, Radware represented that it does not intend to rely on the '801 and '359 patents. See Dkt. No. 453 at 34:7-24. F5 responded that if Radware opens the door to any related patents or applications, including the '702 patent and the application for the '319 patent, then the door should be open for F5 to introduce to all related patents. F5 argues that the impact of PTO proceedings on related patents may be relevant to F5's view of its infringement risk.

If Radware introduces evidence of the '702 patent to show F5's knowledge of the asserted patents, then evidence relating to F5's state of mind with respect to the '702—including whether F5 had reason to believe it was valid, not infringed, or unlikely to be enforced—may be relevant. Because the '801 and '359 patents are also related to the asserted patents, evidence of inequitable conduct in the prosecution of the '801 patent or the PTO's invalidation of the '359 patent may have marginal relevance on the same issues. However, the court is convinced that the likelihood of prejudice and jury confusion outweighs any such relevance. F5 may introduce evidence relating to F5's state of mind only with respect to those patents or applications Radware relies on for willfulness purposes.

VI. F5'S MOTION IN LIMINE NO. 7: TO PRECLUDE USE OF EVIDENCE OF F5 PATENT APPLICATIONS FOR VALIDITY PURPOSES

GRANTED. F5 seeks to preclude Radware from using F5 patent applications to show the nonobviousness of Radware's patents. F5 argues that the distraction, confusion, and prejudice associated with evidence of its own patent applications far outweigh any limited relevance.

Radware argues that F5's Masters application is relevant and probative of nonobviousness. Radware claims that the Masters application and the asserted patents cover the same technology, and Radware asserts that the Masters application is evidence of certain secondary considerations of nonobviousness: commercial success, long felt but unsolved need, the failure of others, and copying. See Dkt. No. 404 at 12. Specifically, Radware argues that the Masters application shows that (1) F5 directed the Masters application at link load balancing technology, thereby by admitting that the technology was novel; and (2) F5 had possession of a LinkProof device and manual before filing the Masters application.

F5 the claims of its patent applications are not co-extensive with the claims of the asserted patents and so show nothing about whether F5 thought Radware's claims were novel. At the pretrial conference, F5 also argued that the Masters application is not relevant to F5's allegations of copying because F5 had developed the Link Controller long before filing the Masters application. Dkt. No. 453 at 69:1-3. F5 also argues that introduction of this evidence shift the jury's focus from the asserted patents and lead to confusion.

F5's motion is granted. The probative value of the Masters application on secondary considerations appears to be slight for the factors cited by Radware—the court particularly questions the application's relevance to commercial success. Presentation of evidence on the claims of F5's applications would likely distract and confuse the jury. Therefore, Radware may not introduce evidence of F5's patents or applications for nonobviousness purposes. VII. F5'S MOTION IN LIMINE NO. 8 AND RADWARE'S MOTION IN LIMINE NO. 2: REFERENCE TO LTM, GTM, AND LINK CONTROLLER AS SEPARATE PRODUCTS FOR INFRINGEMENT AND DAMAGES PURPOSES

GRANTED IN PART AND DENIED IN PART. F5 sells separate licenses to LTM, GTM, and Link Controller—the three software modules on F5's BIG-IP products that Radware accuses of implementing the accused functionality.³ Radware moves to exclude reference to or argument that (1) LTM, GTM, and Link Controller are separate products and that (2) LTM, GTM, and Link Controller modules do not infringe. Dkt. No 380 at 2-5. F5's motion is closely related to Radware's motion. F5 moves: (1) to preclude Radware from presenting any damages⁴ claim based on BIG-IP modules other than LTM, GTM, and Link Controller; and (2) to preclude Radware from offering any damages theory or supporting evidence based on sales of BIG-IP units that include only an LTM module, which, F5 argues, cannot perform any accused functionality unless used simultaneously with F5's GTM module. Dkt. No. 373.

3 LTM, GTM, and Link Controller are three of the 14 total software modules on BIG-IP products. Dkt. No. 399 at 2.

4 Because the parties have agreed that the court, not the jury, should decide any remaining infringement issues, the court finds the portion of F5's motion directed at Radware's infringement theories moot.

As background, Radware argues that the only accused product is F5's BIG-IP hardware, onto which the accused software modules are loaded. In support of its argument, Radware cites this court's October 15, 2015 summary judgment order, which applied the Fantasy Sports and Finjan line of cases⁵ and found that "if the software preloaded onto the BIG-IP products is configured to perform the claimed functions," even though that software is "not accessible until licensed," the BIG-IP products may infringe certain claims.⁶ Dkt. No. 294 at 41. Radware is concerned that referring to the LTM, GTM, and Link Controller as products, as opposed to modules, may mislead the jury into believing that Radware is required to separately prove infringement for modules. Dkt. No. 380 at 4.

5 See Fantasy Sports Properties, Inc. v. Sportsline.com, Inc., 287 F.3d 1108 (Fed. Cir. 2002); Finjan, Inc. v. Secure Computing Corp., 626 F.3d 1197 (Fed. Cir. 2010).

6 Radware also points out that F5 stipulated that the BIG-IP products sold from the start of the applicable damages period until January 2016 infringe claims 1 and 12 of the '319 patent and claims 9, 10, 12, and 14 of the '374 patent. See Dkt. No. 381.

Radware also wishes to limit the impact of this court's summary judgment finding that LTM alone does not infringe claim 24 of the '319 patent. See Dkt. No. 294 at 42. As F5 explains, the only accused functionality with respect to LTM—the capability of using a cost metric for outbound selection as recited in claim 24⁷—can only be accomplished when a user licenses LTM with GTM. Dkt. No. 399 at 3. Based on this finding, F5 moves to preclude Radware from presenting any damages claim based on BIG-IP devices with only an LTM license. Dkt. No. 373 at 1-2. Radware argues that, notwithstanding F5's licensing practices, F5 does not actually sell a product that only contains LTM because all BIG-IP devices contain infringing software at the time of sale, even if that software is not yet licensed and enabled. See Dkt. No. 380 at 4.

7 Radware does not argue that any of its other asserted claims involve LTM. See Dkt. No. 274-17 at 42. The other asserted claims allegedly involve GTM and/or Link Controller. See id. at 45, 47.

At the hearing on the instant motions, Radware raised for the first time an additional theory in support of Radware's position that LTM alone can practice the claimed link load balancing and thus serve as a basis for damages. Specifically, Radware argued that starting around August 2013, BIG-IP LTM-licensed devices included an additional module called "DNS-Lite" that Radware describes as a "try before you buy" version of GTM. Dkt. No. 453 at 9:3-8. Radware thus asserts that even the LTM-only product included a lighter version of GTM that could perform link load balancing. Id. at 9:9-11.

Radware's argument regarding DNS-Lite is untimely, and the court declines to consider it. Despite detailed summary judgment briefing and supporting documentation on lost profits, a motion for reconsideration, motions in limine, and supplemental briefing, to the best of the court's knowledge, none of Radware's damages briefs leading up to the February 11, 2016 pretrial conference even mentioned DNS-Lite. At the pretrial conference, Radware's counsel noted that Radware's expert reports discussed DNS-Lite. Even if the court were to consider Radware's untimely theory, however, the court finds the references to DNS-Lite in the reports of Dr. Rubin (Dkt. No. 179-26 ¶ 102) and Mr. Malackowski (Dkt. No. 274-17 at 44-45) are conclusory and would not allow a reasonable fact finder to evaluate infringement or calculate damages based on DNS-Lite. Radware may have submitted other analysis of DNS-Lite that Radware failed to cite, but it is not the task of the court to scour the record in search of support for a party's arguments. See Keenan v. Allen, 91 F.3d 1275, 1279 (9th Cir. 1996).

F5 argues that Radware's damages theories depend on the fact that LTM, GTM, and Link Controller are individual products that are separately marketed and licensed to F5's customers. Dkt. No. 399 at 2. Radware disputes this characterization and argues that Radware's consistent position has been that the only accused product is "BIG-IP" hardware that includes the accused software modules. Radware argues, for example, that it is only seeking lost profits for complete systems and not for standalone software in the form of "virtual editions" or "add-ons." Dkt. No. 433 at 59:17-21.⁸

8 Radware is seeking a reasonable royalty for virtual editions and add-ons of LTM, GTM, and Link Controller. Dkt. No. 445-6.

The court finds Radware's argument that LTM, GTM, and Link Controller cannot be characterized as separate products unpersuasive for two reasons. First, Radware's expert inconsistently characterizes the LTM, GTM, and Link Controller modules as the "accused products." For example, on page 69 of Mr. Malackowski's report, which Radware cites in opposition to F5's motion in limine, Radware's expert asserts: "Radware lost sales of Alteon and LinkProof devices as a result of F5's sales of the accused BIG-IP Link Controller and GTM and LTM devices." Dkt. No. 274-17 at 69 (emphasis added); see also id. at 39. Second, and more importantly, at no point in Mr. Malackowski's report does he calculate lost profits or a reasonable royalty based on the number of BIG-IP units sold without regard to which modules were licensed on each device. Rather, he separately calculates damages based on the number of BIG-IP devices sold with a license to LTM, the number of BIG-IP devices sold with a license to GTM,⁹ and the number of BIG-IP devices sold with a license to Link Controller.

9 The parties dispute whether Malackowski's figures for GTM include GTM alone or include devices with licenses to both GTM and LTM. Compare Dkt. No. 446 at 11-12 with Dkt. No. 444 at 2-3. As discussed below, this dispute does not affect the court's analysis.

In light of this court's summary judgment analysis, the court rules that F5 may not attempt to argue that pre-hotfix versions of GTM, LTM, or Link Controller do not infringe the asserted claims. The accused product is the BIG-IP device, which has infringed certain claims regardless of which modules were licensed and enabled. However, F5 may argue, for example, that a BIG-IP customer would not be able to use cost-based, outbound link load balancing (or other types of link load balancing) without licenses to particular modules. Because different witnesses have used the terms "products" and "modules" inconsistently, the court will address any violations of the guidelines above on a case-by-case basis.

The court now turns to F5's motion to exclude certain damages arguments. The parties do not appear to have a dispute regarding the first half of F5's motion: Radware agrees with F5 that Radware is not requesting damages pertaining to sales of products that only include licenses for the eleven BIG-IP modules other than LTM, GTM, and Link Controller. See Dkt. No. 405 at 3-4. Accordingly, F5's motion to exclude damages theories and evidence that were not previously disclosed is granted.

F5's motion to exclude damages theories or evidence based on LTM alone requires additional discussion. It is true that the court found that a pre-hotfix BIG-IP device may infringe claim 24 of the '319 patent even without a license that would enable GTM. Radware's opposition to F5's motion concedes, however, that "[i]nfringement is an entirely separate inquiry from damages." Dkt. No. 405 at 3. Notwithstanding the court's infringement finding, the court has found that to actually use cost-based, outbound link load balancing, a customer needs licenses to enable both LTM and GTM. See Dkt. No. 294 at 42. Thus, Radware has no basis to argue that the buyer of a BIG-IP unit with only a license to LTM (and not GTM) purchased that device to utilize cost-based, outbound link load balancing. Put differently, Radware cannot show that but for LTM's ability to perform cost-based, outbound link load balancing (which LTM cannot do without GTM), a BIG-IP customer with only a license to LTM would have bought from Radware.

Moreover, as explained above, Radware's expert based his damages calculations on the number of BIG-IP units with LTM, GTM, and/or Link Controller licensed and enabled. Radware's expert has admitted that his report assumed that LTM could perform outbound link load balancing on its own. Dkt. No. 373-9 at 20:15-19. Mr. Malackowski also testified that "to the extent the jury determines that [LTM] does not infringe, then it would be removed from my analysis." Id. at 70:17-19.

A court may preclude a plaintiff from introducing new damages theories at trial. See, e.g., Oracle USA, Inc. v. SAP AG, 264 F.R.D. 541, 557 (N.D. Cal. 2009). The court finds that precluding Radware and its expert from claiming damages for BIG-IP units with LTM alone is appropriate in this case. The court need not address Radware's assertion that it was "entitled to seek damages for sales of all accused BIG-IP products" because Radware admits that it "limited its requested damages award to sales of BIG-IP products where one or more of the LTM, GTM, or Link Controller modules were licensed." Dkt. No. 405 at 1. Radware cannot credibly claim that it disclosed a damages theory suggesting that Radware should receive lost profits or a reasonable royalty based on the number of BIG-IP units sold without regard to which modules were licensed and enabled on each device. The implicit assumption in Radware's expert disclosures is that Radware is only entitled to damages for BIG-IP units that could actually perform the accused forms of link load balancing. Because LTM alone cannot perform cost-based, outbound link load balancing, Mr. Malackowski's methodology requires him to exclude damages calculations aimed at LTM alone.

To the extent that Radware argues that the court's ruling deprives Radware of damages adequate to compensate for admitted or adjudged infringement, the court notes that its ruling does not preclude Radware from seeking lost profits or a reasonable royalty for BIG-IP units in which both GTM and LTM were licensed and enabled.¹⁰ Radware asserts that the sales figures Mr. Malackowski analyzed under the heading of "GTM" actually included units in which both GTM and LTM were licensed. Dkt. No. 446 at 12.¹¹ Accordingly, excluding sales of devices with only a license to LTM does not preclude Radware from obtaining damages, consistent with Mr. Malackowski's opinions, for the BIG-IP units in which customers could actually use the functionality described in claim 24 of the '319 patent and the other asserted claims.

10 Section 284, of course, provides that a patentee is entitled to an award not less than a reasonable royalty for infringement. Radware acknowledges, however, that infringement and damages are separate inquires. Here, Radware's proffered damages theory incorrectly assumes that LTM alone can provide outbound link load balancing capability and does not explain how to calculate a reasonable royalty for LTM if the assumption that LTM alone provides outbound link load balancing capability proves false. See Dkt. No. 274-17 (Malackowski Rep.) at 109.

11 F5 disputes Radware's assertion, Dkt. No. 444 at 2-3, but any factual dispute regarding the content of the figures can be resolved at trial.

VIII. RADWARE'S MOTION IN LIMINE NO. 1: TO PRECLUDE ARGUMENT REGARDING RADWARE'S ALLEGED IMPROPER COMPETITIVE ACTS

GRANTED. In support of Radware's willfulness theory, Radware will seek to introduce evidence and testimony that F5 copied Radware's patented products. Radware moves under Federal Rules of Evidence 403 and 404(b) to preclude arguments regarding its own alleged improper competitive acts, such as arguments that Radware also purportedly possesses competitor documents or copies competitor products, as irrelevant to any claim or defense in this action. Dkt. No. 380 at 1-2. F5 responds that F5 should be able to point out that Radware has engaged in the same conduct to show that there is a standard industry practice of monitoring competitive products and literature. Dkt. No. 399 at 1. The court agrees with F5 that F5 may show industry practice, but the court also agrees with Radware that any marginal relevance of Radware's alleged bad acts would be greatly outweighed by the associated unfair prejudice, waste of time, and confusion. This order does not preclude F5 from showing that it is a common industry practice to monitor competitors' products and patents, to the extent relevant. This order also does not preclude F5 from arguing that Radware's patents are invalid in light of the prior art.

IX. RADWARE'S MOTION IN LIMINE NO. 3: TO PRECLUDE ARGUMENT THAT GSLB IS LLB

DENIED. Radware moves on the basis of judicial estoppel to preclude any argument, reference, or claim that Global Server Load Balancing ("GSLB") is the same as Link Load Balancing ("LLB") or that any GSLB product or reference, such as such as the Cisco DistributedDirector ("Cisco DD") and F5's own 3DNS, invalidates Radware's asserted claims. Dkt. No. 380 at 5-8. Radware argues that during claim construction in this case, F5 proposed a narrow construction of the term "proximit[y]/[ies]," and F5 stated that "the claims of the asserted patents have nothing to do with multiple server farm environments (i.e., GSLB)." Id. at 5-6 (citing Dkt. 156 of Related Case No. 5:13-cv-02021-RMW at 11). According to Radware, F5's present position—that specific references disclosing or practicing GSLB anticipate the asserted claims—is clearly inconsistent with F5's prior position. Id. at 7. Moreover, Radware argues, the court adopted a narrower construction of "proximit[y]/[ies]" at F5's request, and F5 would derive an unfair advantage from being able to change its position now. Id. at 7-8.

F5 responds that judicial estoppel does not apply. Dkt. No. 399 at 4-8. First, F5 clarifies that it is not using GSLB in general as a prior art reference.¹² Rather, F5 relies on specific embodiments of GSLB, such as the Cisco DD. Id. at 5. Moreover, F5 argues that its positions are not inconsistent. F5 notes that its prior statement focused on multiple server farm environments, not simply GSLB. Id. at 6. F5 also argues that its claim construction position has not changed:

12 Indeed, this court's summary judgment order precluded F5 from relying on GSLB in the abstract because F5 did not timely disclose GSLB as a prior art reference. Dkt. No. 294 at 9 n.3.

F5 acknowledges (as it did during claim construction) that the asserted claims relate to link load balancing and multi-homed environments. It is simply also arguing that the asserted claims—construed to cover the link load balancing embodiments of the specification—are invalidated by GSLB products that were capable of performing the recited aspect of link load balancing.

Id. at 7. Moreover, F5 argues, the court disagreed with both parties' constructions,¹³ and there is no prejudice to Radware because Radware has long been on notice of F5's prior art reference. Id. at 8.

13 F5 also points out that Radware is no longer asserting any claims that recite "proximit[y]/[ies]."

The court finds that judicial estoppel does not apply. On the one hand, F5 previously argued that the term "network proximity," which appears in the asserted patents' specifications and is associated with GSLB, should not dictate the meaning of the term "proximity" in certain asserted claims. The reason is that the patents' specifications distinguish "proximity" from "network proximity." See Dkt. No. 122 at 7. On the other hand, F5 now argues that specific prior art products with GSLB functionality could disclose every limitation of particular claims that do not even recite the term "proximity." These arguments are not "clearly inconsistent" even if, in the context of construing "proximit[y]/[ies]," F5 may have made some statements that were broader than necessary for claim construction. The court also finds that there is no prejudice to Radware because Radware has been aware of F5's planned prior art references for months, and Radware has had adequate opportunity to defend against them. Accordingly, Radware's motion is denied.

X. RADWARE'S MOTION IN LIMINE NO. 4: TO EXCLUDE REFERENCE TO THE FACT THAT RADWARE HAS NOT YET SUED OTHER COMPANIES THAT OFFER INFRINGING ALTERNATIVES

GRANTED. In support of Radware's argument that there are no acceptable non-infringing alternatives to its patented products, Radware may offer testimony suggesting that particular competitors' products infringe. Radware moves under FRE 403 to exclude any testimony from F5 relating to the fact that Radware has not yet brought suit against other competitors. Dkt. No. 380 at 8-10. Specifically, Radware concedes that "F5 is free to argue that acceptable non-infringing alternatives existed on the market," id. at 8, but Radware argues that whether Radware has filed suit is, at most, evidence of whether Radware believes that its competitors' products infringe, not whether they actually infringe, id. at 9.

F5 responds by arguing that the jury should be able to hear all of relevant information to determine whether Radware has proven that such other products infringe. Dkt. No. 399 at 9. According to F5, this includes the fact that Radware has not sued such other parties. Id. F5 notes that to the extent that there is a concern that the jury will be misled, the court could instruct the jury that there is no obligation to sue all potential infringers at once. Id.

The court finds that because there are many factors unrelated to the merits of a case that go into whether a lawsuit should be filed, the question of whether Radware has sued other companies is only minimally relevant to the question of whether these companies' products are acceptable non-infringing alternatives, and the risk of unfair prejudice is significant. Accordingly, Radware's motion is granted. The court may be willing to revisit its ruling if, at trial, F5 is able to point to particular arguments or evidence raised by Radware that can only fairly be addressed by reference to the lack of a lawsuit.

XI. RADWARE'S MOTION IN LIMINE NO. 5: TO EXCLUDE EVIDENCE AND ARGUMENT REGARDING THE NUMBER OF DOWNLOADS OF THE F5 PATCH

DENIED. Radware moves to exclude evidence and argument regarding the number of customers who have downloaded F5's patch to reactivate certain link load balancing functionality in the accused BIG-IP product. Dkt. No. 380 at 11-14. By way of background, after Radware filed the instant lawsuit, F5 made changes to the software on the accused products via two "hotfixes," purportedly to remove the functionality Radware accuses of infringement. If F5 customers want to restore the functionality that was lost with the first hotfix,¹⁴ they can download a software patch from F5. Dkt. No. 303-2 ¶ 4. The patch, however, is only available to customers outside the United States, id., and very few foreign customers have requested the patch, Dkt. No. 303 at 10 n.4. F5 wants to use that fact to show a lack of demand for Radware's inventions. Radware makes two principal arguments for excluding this evidence.

14 It is not clear from the present record whether the patch affects the second hotfix.

First, Radware argues, the number of downloads is irrelevant to infringement because the hotfix and patch have no effect on the functionality covered by six of the seven asserted claims. Dkt. No. 380 at 13. Moreover, F5 has stipulated to infringement prior to January 1, 2016 of the aforementioned six claims: claims 1 and 12 of the '319 Patent and claims 9, 10, 12, and 14 of the '374 Patent. Dkt. No. 381. F5 does not dispute these points. See Dkt. No. 399 at 11 (arguing that the lack of patch downloads is relevant "regardless of the fact that Radware no longer asserts claims covering most of the specific types of LLB that were the subject of the first Hotfix"). Thus it appears undisputed that the first hotfix was not aimed at preventing infringement of claims 1 and 12 of the '319 Patent or claims 9, 10, 12, and 14 of the '374 Patent and that the patch did not affect the functionality that Radware accuses of infringing these claims. Therefore, the court agrees with Radware that F5 will not be allowed to discuss demand for the patch in reference to these claims. Any such discussion would be irrelevant.

If not for Radware's assertion of claim 24 of the '319 patent, the court would agree that the patch has no relevance. As F5 points out, however, the patch did affect the functionality that Radware still accuses of infringing claim 24 of the '319 patent. Id. The court granted summary judgment of infringement of claim 24 of the '319 patent for products sold before F5 implemented its first hotfix, Dkt. No. 294 at 43, but the court also granted summary judgment of non-infringement of the same claim for products with the hotfix in place, Dkt. No. 420. Radware still seeks damages for infringement of claim 24 by F5 products sold before the hotfix was in place. Therefore, the court finds that demand for patch, which can restore the functionality covered by claim 24, could be relevant to demand for the patented features and to damages. Radware's motion does not separately address claim 24.

While it is true that an F5 witness submitted a declaration indicating that the patch is not available in the United States, Dkt. No. 303-2 ¶ 4 (Brewer Decl.), F5 points out that Radware bases its damages model in part on F5 products made in the United States and sold abroad. To the extent that Radware argues that foreign demand for the accused products is relevant, F5 should have the opportunity to refute Radware's arguments. Radware argues that it does not have a reliable baseline on the number of accused BIG-IP units on the market to which the patch could even be applied. See Dkt. No. 380 at 13. The court finds, however, that any issues with the reliability of using the number of patch downloads to measure demand go to the weight of the evidence, not to its admissibility.

XII. RADWARE'S MOTION IN LIMINE NO. 6: TO PRECLUDE ARGUMENT THAT DEMAND FOR THE PATENTED PRODUCT(S) OUTSIDE THE U.S. IS IRRELEVANT

GRANTED AS UNOPPOSED. Radware moves to preclude argument that because demand for link load balancing in the United States is minimal compared to demand in foreign markets, Radware is not entitled to any damages. Dkt. No. 380 at 14-16. F5 responds that it does not plan to argue that foreign demand is irrelevant. Dkt. No. 399 at 13-14. Rather, F5 plans to argue that foreign-manufactured products that satisfy foreign demand constitute non-infringing alternatives and thus undermine Radware's claims for lost profits. Id. at 13. Because F5 does not dispute the relevance of foreign demand, Radware's motion is granted as unopposed.

XIII. RADWARE'S MOTION IN LIMINE NO. 7: TO PRECLUDE KEVIN DELGADILLO FROM TESTIFYING AS AN EXPERT WITNESS

GRANTED IN PART AND DENIED IN PART. Radware moves to exclude Kevin Delgadillo, a former Cisco product manager and the author of several documents related to prior art Cisco technology, from testifying as a technical expert witness. Dkt. No. 380 at 16-23. It is undisputed that F5 did not disclose Delgadillo as an expert in this case.

Radware first questions Delgadillo's qualifications to testify about Cisco technology by noting that he is not an engineer and that he holds an MBA. Id. at 16-19, 22-23. Moreover, Radware points out that during a deposition, Delgadillo admitted: "I do not have the technical expertise on this that I used to." Id. at 22 (citing Dkt. No. 380-9 at 36:2-21).¹⁵

15 The court notes that while Radware's motion cites twenty lines of Delgadillo's testimony, it provides only an excerpt two lines of testimony without the surrounding context. The surrounding lines suggest that Delgadillo has personal knowledge of the subjects of his testimony.

In response, F5 clarifies that F5 is only trying to call Delgadillo as a fact witness who is anticipated to "testify about prior art references and systems related to the Cisco Distributed Director and Local Director products . . . and [t]o the extent his deposition testimony included other topics, he might also address those." Dkt. No. 399 at 15. In any event, F5 argues, Delgadillo is qualified to discuss Cisco products. F5 points out that Delgadillo has a bachelor's degree in applied physics from the California Institute of Technology. Id. at 17. Moreover, Delgadillo testified that in his role as a product manager, he understood both the technical and business aspects of the products in question. Id. at 18 (citing Dkt. No. 399-3 at 12:4-8). F5 thus argues that Delgadillo should be able to testify regarding his personal knowledge of Cisco prior art technology.

Radware also argues that Delgadillo's testimony "appears to be based upon" hearsay based on certain responses Delgadillo gave during his deposition:

Q: Do you have any knowledge as to whether multihoming is common today across enterprises? . . .
A: I would say yes, it's probably common.
Q. What's that based on?
A. Just a guess, it's an educated guess.

Dkt. No. 380-9 at 50:14-22; see also id. at 13:4-20, 18-22. Radware thus seeks to preclude Delgadillo from testifying about certain trial exhibits listed in Radware's motion. F5 responds that Radware's concern about hearsay is speculative and that F5 does not intend for Delgadillo to provide improper hearsay testimony. Dkt. No. 399 at 18.

The court concludes that as long as F5 provides sufficient foundation, Delgadillo may testify as to facts about which he has personal knowledge. Radware concedes as much. See Dkt. No. 380 at 19 (indicating that Delgadillo "can testify about facts regarding which he has personal knowledge, such as, for example, the date on which he wrote the DistributedDirector White Paper, as a lay witness"). Delgadillo may not offer opinions as an expert witness. Contrary to Radware's apparent position, however, even a fact witness such as Delgadillo may draw upon technical knowledge. As this court explained in a prior case, "there are limits to what constitutes 'expert' testimony subject to advance disclosure, written reports, Daubert gatekeeping, and so on. If there were not, patent litigation would be more astronomically expensive than it already is." Hynix Semiconductor Inc. v. Rambus Inc., No. C-00-20905 RMW, 2009 WL 230039, at *10 (N.D. Cal. Jan. 27, 2009). To the extent that Radware objects to specific questions or exhibits, Radware may raise those objections at trial.

XIV. RADWARE'S MOTION IN LIMINE NO. 8: TO EXCLUDE REFERENCES TO F5'S COUNTERCLAIMS

GRANTED AS UNOPPOSED. Radware seeks to exclude any reference to F5's asserted counterclaims at the trial on Radware's claims for patent infringement. Dkt. No. 380 at 23-25. Radware notes that a separate trial will be held on those claims. F5 does not dispute Radware's motion.

XV. RADWARE'S MOTION IN LIMINE NO. 10: TO EXCLUDE EVIDENCE AND ARGUMENT REGARDING THE FACT THAT THERE WAS A PRIOR LAWSUIT BETWEEN F5 AND RADWARE

GRANTED IN PART. Radware moves to exclude any evidence to the prior litigation between the parties—specifically seeking to preclude reference to F5 Networks, Inc. v. Array Networks, Inc. et al., Civil Action No. 2:03-cv- 00688-MJP (W.D. Wash). Radware argues that the Washington case is not relevant and that any reference to the Washington case may suggest to the jury that Radware is motivated in this lawsuit by monetary gain or revenge. F5 opposes the motion, asserting that settlement of the Washington case resulted in a license agreement. See Dkt. No. 397 at 2. The damages experts for both parties discuss the license agreement in the reasonable royalty sections of their reports.

In its motion, Radware suggests that the license agreement did not arise out of prior litigation between the parties. See Dkt. No. 378-6 ("around the same time, Radware and F5 entered into a non-exclusive patent license agreement"). However, Radware's damages expert Mr. Malackowski indicates in his report that the license agreement "appear[s] to be related to the settlement of litigation." Dkt. No. 247-17 ¶15.1.11. Mr. Malackowski cites the Washington case in support of this statement. Id. at n. 564. F5's expert, Ms. Stamm, discusses the license agreement in her report, but she does not suggest that it was the result of settlement. Dkt. No. 204-27 ¶¶131-137. The parties' briefs and expert reports do not establish whether the license agreement arose out of the Washington settlement.

To the extent that either party seeks to present evidence concerning the license agreement as relevant to a reasonably royalty determination, the court will need to assess the admissibility of the license agreement for that purpose. As part of the admissibility inquiry, the court will consider the context in which the license agreement arose, including whether it arose as part of litigation settlement. Therefore, the court reserves judgment on whether evidence of the allegedly underlying litigation is relevant or admissible to give context to the license agreement.

Radware's motion is otherwise granted as unopposed.

XVI. RADWARE'S MOTION IN LIMINE NO. 12: TO EXCLUDE EVIDENCE AND ARGUMENT OF PRIOR DAUBERT CHALLENGES AND EXCLUSIONS TO THE TESTIMONY OF JAMES MALACKOWSKI ON UNRELATED MATTERS

GRANTED. Radware seeks to exclude any reference to prior Daubert challenges to or exclusions of the opinions of Radware's damages expert, James Malackowski. The court agrees that although Mr. Malackowski's prior experience as an expert witness is relevant, the danger of unfair prejudice outweighs the probative value of such evidence.

F5 argues that if Radware's motion is granted, it should be granted with respect to all expert witnesses. F5 also argues that if Radware "opens the door" by referring to Mr. Malackowski's prior testimonial experience to bolster Mr. Malackowski's expert credentials, F5 should be permitted to introduce evidence showing that courts have rejected such testimony.

Radware's motion is granted. The court will exclude evidence of prior Daubert challenges for all expert witnesses. However, to the extent either party introduces evidence of the prior testimonial experience of any expert to imply that the expert's opinion were blessed by the court, the court may permit cross-examination on the outcome of such testimony.

XVII. RADWARE'S MOTION IN LIMINE NO. 13: TO EXCLUDE EVIDENCE OR ARGUMENT REGARDING THE WEALTH OF MR. ZISAPEL AND/OR THE ZISAPEL FAMILY

MOOTED BY STIPULATION. See Dkt. Nos. 435.

XVIII. RADWARE'S MOTION IN LIMINE NO. 15: TO EXCLUDE EVIDENCE AND TESTIMONY REGARDING RADWARE'S OVERALL SIZE OR LITIGATION EXPENDITURES

WITHDRAWN BY STIPULATION. See Dkt. Nos. 435. XIX. RADWARE'S MOTION IN LIMINE NO. 16: TO EXCLUDE REFERENCE, EVIDENCE OR TESTIMONY REGARDING RADWARE'S NARROWING OF CLAIMS FOR TRIAL

GRANTED. Radware seeks to exclude any reference to Radware's narrowing of claims for trial. F5 opposes this motion, arguing that it should be able to introduce evidence that the first hotfix avoided infringement of several claims no longer asserted by Radware. F5 claims such evidence is relevant to the willfulness inquiry because it shows that F5 did not act recklessly. The court is not convinced that Radware's decision to drop certain claims has any relevance to F5's alleged willful infringement of the remaining asserted claims. Radware's motion is granted.

XX. RADWARE'S DAUBERT MOTIONS/MOTIONS TO EXCLUDE EXPERT TESTIMONY

DENIED. Pursuant to Federal Rules of Evidence 401, 402, 403, 702, and the principles of Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993), Radware seeks (1) to limit the testimony of Dr. Peter Alexander, F5's technical expert; (2) to limit the testimony of Ms. Laura Stamm, F5's damages expert; and (3) to exclude in its entirety the testimony of Mr. Ethan Banks, F5's expert on networking. F5's experts may testify as to the opinions expressed in their reports.

A. Dr. Alexander

First, Radware argues that Dr. Alexander's testimony should be limited to the opinions disclosed in his report. See Dkt. No. 377 at 4. F5 represents that it does not intend to have Dr. Alexander offer any opinions outside the scope of his report. The substance of Radware's motion appears to focus on Dr. Alexander's inability to testify substantively on certain issues at deposition—not whether he disclosed the opinions in his report. To the extent Radware's motion addresses Dr. Alexander's noninfringement opinions, it is moot.

Radware lists several additional topics on which Radware claims that Dr. Alexander was unable to testify substantively at deposition. It is not clear that Dr. Alexander ever offered an opinion on these topics. For example, Radware seeks to preclude Dr. Alexander's testimony on whether the claims of the Radware patents-in-suit cover BGP. F5 notes that Dr. Alexander discusses BGP in his report, but F5 does not suggest that Dr. Alexander will offer an opinion on whether the asserted patents cover BGP. The court intends to hold Dr. Alexander to the disclosures included in his report at trial. F5 confirmed at the pretrial conference that Dr. Alexander will not offer opinions outside the scope of his report. Dkt. No. 453 at 31:20-24. Accordingly, Radware's motion is denied as moot. If Dr. Alexander's testimony exceeds the scope of his report, Radware may object at trial.

Second, Radware seeks to preclude Dr. Alexander from opining on whether Dr. Rubin's infringement analysis is legally sufficient. Third, Radware seeks to exclude any testimony from Dr. Alexander on the contents of his January 11, 2016 declaration. This declaration addresses F5's alleged infringement after the first and second hotfixes. Because the parties have agreed that the court will determine the only remaining infringement issue, Radware's motion on these topics is moot.

B. Ms. Stamm

Radware seeks to exclude certain opinions expressed in Ms. Stamm's March 2, 2015 expert report, arguing that the opinions are based on an erroneous understanding of the law and unreliable methodology.

1. Application of Panduit Factors

Radware argues that Ms. Stamm's application of the Panduit factors is inconsistent with Federal Circuit law. Radware asserts that Ms. Stamm incorrectly interpreted Panduit factor one to require a showing that the patented feature is a driver of demand. Radware is correct that under Federal Circuit law, Panduit factor one "simply asks whether demand existed for the patented product"—not whether demand existed for the patented feature. DePuy Spine, Inc. v. Medtronic Sofamor Danek, Inc., 567 F.3d 1314, 1330 (Fed. Cir. 2009). However, demand for the patented feature may still be relevant to other Panduit factors or the "but for" analysis as a whole. See, e.g., DePuy Spine, 567 F.3d at 1330-31("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").

Having reviewed the relevant paragraphs of Ms. Stamm's report, the court finds that Ms. Stamm's opinion on demand for the patented feature is not limited to Panduit factor one. See Dkt. No. 204-27 ¶ 46 ("Based on my review of the evidence Radware cannot establish lost profits because it cannot establish that the benefits of the patents-in-suit were a driver of consumer demand such that the inclusion of the patented technology in these products led to loss of sales of Radware's products." ). In contrast, Ms. Stamm clearly limits other opinions to a particular factor. Id. ¶ 46 ("Furthermore Mr. Malackowski has not adequately established that the second Panduit factor is satisfied."). Radware's motion to exclude Ms. Stamm's lost profits opinion for failure to apply the correct legal standard is denied.

2. Opinion on Commercially Acceptable Non-infringing Alternatives

Radware also moves to exclude Ms. Stamm's opinion on commercially acceptable non-infringing alternatives. Radware objects that Ms. Stamm's report contains no discussion about whether BGP would be an acceptable non-infringing alternative. Radware further objects that Ms. Stamm's report contains no discussion of a GTM or LTM product with link load balancing functionality removed as an alternative.

With respect to BGP, Radware first argues that Ms. Stamm opined only on the availabilitity of alternatives, not whether such alternatives would be commercial acceptable.

In her report, Ms. Stamm relies on Mr. Banks for her understanding that BGP is a commercially acceptable alternative: "I understand from Mr. Banks that BGP is a protocol that has long existed, and that large enterprises with a business requirement for multi-homing more commonly use BGP than appliances for link load balancing." Dkt. No. 204-27 at ¶ 79. It is appropriate for Ms. Stamm to rely on the opinions of other experts.

Ms. Stamm also relies on evidence of BGP's commercial acceptability from F5 employees. See id. ¶ 79 ("According to document prepared by Mr. Needham, the strength of BGP is that it is widely accepted and used on most routers but a weakness is that it is difficult to manage"); ("Mr. Schaller testified that he is aware that high end enterprises with multiple locations use BGP as a solution to balance load.") (emphases added). Ms. Stamm may rely on factual evidence supplied by Mr. Needham and Mr. Schaller if F5 presents such foundation evidence at trial. See Oracle Am., Inc. v. Google Inc., No. C 10-03561 WHA, 2011 WL 5914033, at *1 (N.D. Cal. Nov. 28, 2011) ("Expert reliance on foundational facts supplied by [defendant's] engineers can be proper so long as they testify to the foundational facts with firsthand knowledge.") (citing Therasense, Inc. v. Becton, Dickinson & Co., No. C 04-02123 WHA, 2008 WL 2323856 at *2 (N.D. Cal. May 22, 2008) (Alsup, J.) ("The traditional and correct way to proceed is for a foundational witness to testify first-hand at trial to the foundational fact . . . and to be cross-examined. Then the expert can offer his or her opinion on the assumption that the foundational fact is accepted by the jury.")).

Radware argues that Ms. Stamm did not perform any analysis of the acceptability of the non-infringing alternative to consumers, citing Ms. Stamm's deposition testimony. This is not a basis to exclude Ms. Stamm's testimony on BGP as an alternative; it is not clear that Ms. Stamm would have the expertise to offer such an opinion in any case. F5 will be able to present the foundational evidence for Ms. Stamm's opinions on non-infringing alternatives, and Radware will have the opportunity to test that evidence.

Radware also argues that Ms. Stamm did not discuss in her report the possibility that a hypothetical GTM or LTM with link load balancing functionality removed would be an acceptable non-infringing alternative.¹⁶ However, a review of Ms. Stamm's report shows that it includes a section entitled "The Patented Features Are Rarely Used and Could Be Removed Without Affecting Demand." See Dkt. No. 204-27 at 25. Although this section is not included within the heading "There were Acceptable Non-Infringing Substitutes Available During the Damages Period" (id. at 30), the court finds the disclosure in the report to be clear enough to put Radware on notice of Ms. Stamm's opinions.

16 Radware also cites Siemens Medical Solutions v. Saint-Gobain Ceramics, 637 F. 3d 1269, 1289 (Fed. Cir. 2011) to argue that GTM and LTM without link load balancing cannot constitute viable alternatives for lost profits purposes because such products do not exist. See Dkt. No. 377 at 14 (citing 637 F. 3d 1269 at 1289 ("if the substitute cannot be commercialized 'readily,' then it is not available for purposes of a lost profits determination.")). However, as the Siemens court acknowledges, a "substitute need not be on sale at the time of infringement." Id. at 1289. "When an alleged alternative is not on the market during the accounting period, a trial court may reasonably infer that it was not available as a noninfringing substitute at that time." Grain Processing Corp. v. Am. Maize-Prods. Co., 185 F.3d 1341, 1353 (Fed. Cir.1999). The burden then shifts to the alleged infringer to show that the substitute was "available" based on alternative actions the alleged infringer reasonably could have taken. DePuy Spine, 567 F.3d at 1331 (Fed. Cir. 2009). If the accused infringer had the necessary materials, equipment, know-how, and experience to make an alternative product during the relevant time frame, the product may have been "available." Grain Processing Corp., 185 F.3d at 1353. To the extent such opinions are disclosed in her report, Ms. Stamm may opine on the "availability" of LTM and GTM without the accused functionality. --------

3. Reliance on QK View Study

Radware also seeks to preclude Ms. Stamm from relying on F5's QK View study to show absence of demand for link load balancing. Radware objects to the QK View study as unreliable because it represents only a "snapshot" of F5 customer's configuration files that come only from F5 customers who voluntarily submit the data as part of a request for service. Radware further objects to Ms. Stamm's reliance on the study to conclude that GTM and LTM customers rarely use the patented functionality; Radware argues that her analysis is logically unsound because the study is an under-inclusive survey of an unknown number of customers. F5 argues that not only is the survey reliable, but that Ms. Stamm's reliance on the study is permissible under Federal Rule of Evidence 703 because it is "of a type reasonably relied upon by experts in the particular field in forming opinion or inferences on the subject." Dkt. No. 390 at 18.

The Ninth Circuit recognizes that "survey evidence should ordinarily be found sufficiently reliable under Daubert. Unlike novel scientific theories, a jury should be able to determine whether asserted technical deficiencies undermine a survey's probative value." Southland Sod Farms v. Stover Seed Co., 108 F.3d 1134, 1143 n.8 (9th Cir.1997) (internal quotation marks and citations omitted); see also Clicks Billiards, Inc. v. Sixshooters, Inc., 251 F.3d 1252, 1262 (9th Cir. 2001) ("Technical unreliability goes to the weight accorded a survey, not its admissibility."). "Treatment of surveys is a two-step process. First, is the survey admissible? That is, is there a proper foundation for admissibility, and is it relevant and conducted according to accepted principles? This threshold question may be determined by the judge. Once the survey is admitted, however, follow-on issues of methodology, survey design, . . . the experience and reputation of the expert, critique of conclusions, and the like go to the weight of the survey rather than its admissibility." Clicks, 251 F.3d at 1263 (internal citations omitted).

F5 maintains a repository of QK View files, which are diagnostic "snapshots" of a machine at a particular time when someone requested information on how to fix problems related to an F5 product. F5 asserts that these files provide relevant data as to whether customers had configurations supporting link load balancing. Radware objects that the data is unreliable because customers submit the files voluntarily, and files can be deleted on customer request. Radware also points out that the files are not necessarily updated if customers change their configurations. Radware also asserts that the QK View repository may be replete with errors and duplicates because the F5 employee who compiled the data testified that F5 sometimes gets "the same QK view three or four times or more often." Dkt. No. 376-4 at 14-16.

Although the court agrees that the QK View study presents certain reliability issues, the jury will be able to weigh the probative value of the study. The court notes that F5 represents that Radware will have the opportunity to examine both Mr. Brain, an F5 employee with knowledge of the QK View study, and Ms. Stamm at trial. See Dkt. No. 390 at 18. Although the QK View study may well be under-inclusive, this does not justify excluding Ms. Stamm's testimony. See Fujifilm Corp. v. Motorola Mobility LLC, No. 12-CV-03587-WHO, 2015 WL 1737951, at *6 (N.D. Cal. Apr. 8, 2015) ("That the Imaging Feature Study questioned only 'mid/high tier smartphone owners' does not justify excluding Pardy's testimony."); see also Smartflash LLC v. Apple, Inc., No. 13-cv-00447, 2014 WL 7336213, at *4 (E.D. Tex. Dec. 23, 2014) (declining to exclude survey evidence of "only 'regular users' instead of all 'purchasers' of accused devices" because methodology goes to " the weight of the survey evidence, not its admissibility").

Radware cites to General Electric Co. v. Joiner, 522 U.S. 136 (1997) to argue that Ms. Stamm's opinion is not sufficiently tied to the data on which she relies. In Joiner, however, the Court found that the "studies were so dissimilar to the facts presented in this litigation that it was not an abuse of discretion for the District Court to have rejected the experts' reliance on them." 522 U.S. at 144-45. The QK View files do not suffer from this problem—the files do show the F5 customers' configurations, whereas the studies in Joiner were "seemingly far-removed animal studies" involving different levels and types of exposure to a carcinogen than were at issue in the case.

Radware's motion to exclude Ms. Stamm's testimony on the QK View study is denied.

C. Mr. Banks

Radware seeks to prevent Mr. Banks from opining that (1) border gateway protocol ("BGP") and global sever load balancing ("GSLB") are acceptable alternatives to link load balancing and (2) that customers would utilize BGP and GSLB instead of link load balancing.

1. Qualifications

Radware argues that Mr. Banks does not have the requisite "scientific, technical, or other specialized knowledge will assist the trier of fact" to opine that BGP or GSLB are alternatives to link load balancing. Radware argues that he has only limited work experience and no personal experience with link load balancing. F5 responds that Mr. Banks has sufficient credentials to establish him as a networking expert, citing his 20 years of experience in the networking field, including server load balancing, global server load balancing, and ISP load balancing.

The parties do not appear to dispute that Mr. Banks is a qualified expert in networking. Rather, Radware challenges the basis for his specific opinions on alternatives to link load balancing, arguing that he performed no study or analysis of link load balancing. Having reviewed Mr. Banks' credentials, the court is satisfied that he has sufficient industry experience to offer opinions on ISP load balancing functionalities and how enterprises use such functionalities.

2. Reliability

Radware also challenges the reliability of Mr. Banks' opinions, arguing that he employs no discernable methodology. F5 argues that Mr. Banks' opinion on how BGP can accomplish load balancing across ISPs in a multi-homed environment is testable under Daubert, and that the balance of his opinions are based on personal observations and experience—which F5 argues is permissible under Federal Rule of Evidence 702. See Dkt No. 390 at 23-24 (citing Pipitone v. Biomatrix, Inc., 288 F.3d 239 (5th Cir. 2002)).

Federal Rule of Evidence 702 does not require verification, only "reliable principles and methods." The Supreme Court clearly stated that "no one denies that an expert might draw a conclusion from a set of observations based on extensive and specialized experience." Kumho Tire Co. v. Carmichael, 526 U.S. 137, 156 (1999); see also Fed. R. Evid. 702, advisory committee's note ("Nothing in this amendment is intended to suggest that experience alone—or experience in conjunction with other knowledge, skill, training or education—may not provide a sufficient foundation for expert testimony."). The advisory committee to Rule 702 also notes that "[s]ome types of expert testimony will be more objectively verifiable, and subject to the expectations of falsifiability, peer review, and publication, than others. Some types of expert testimony will not rely on anything like a scientific method, and so will have to be evaluated by reference to other standard principles attendant to the particular area of expertise. The trial judge in all cases of proffered expert testimony must find that it is properly grounded, well-reasoned, and not speculative before it can be admitted."

In this case, the court finds that Mr. Banks bases his opinions on his experience in the networking industry, and that he explains how he reached his conclusions. See, e.g., Dkt. No. 377-7 ¶ 9 ("In my experience managing large enterprise networks for higher education, financial services, state and local governments, technology corporations, and medical organizations, I have never been given a business requirement that necessitated a link load balancing appliance to meet business needs."). Radware's motion to exclude Mr. Banks' opinions as unreliable is denied. Radware's challenges to Mr. Banks' opinions go to weight and may be developed on cross-examination.

IT IS SO ORDERED. Dated: February 13, 2016

/s/_________

Ronald M. Whyte

United States District Judge