From Casetext: Smarter Legal Research

Radware, Ltd. v. F5 Networks, Inc.

UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA SAN JOSE DIVISION
Oct 31, 2016
Case No. 5:13-cv-02024-RMW (N.D. Cal. Oct. 31, 2016)

Opinion

Case No. 5:13-cv-02024-RMW

10-31-2016

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


ORDER DENYING F5'S MOTION FOR NEW TRIAL

Re: Dkt. No. 624

A jury found that the devices that defendant F5 Networks Inc. had argued anticipated the asserted claims of plaintiff Radware's U.S. Patent Nos. 8,266,319 ("'319 Patent") and 8,484,374 ("'374 Patent") were not actually prior art to Radware's asserted patents. Before the court is F5's motion for a new trial on the issue of invalidity. Dkt. No. 624. The court finds this motion suitable for decision without oral argument pursuant to Civil Local Rule 7-1(b) and thus vacates the hearing set for November 18, 2016. Having considered the parties' submissions, the relevant law, and the record in this case, the court DENIES F5's motion for a new trial.

I. BACKGROUND

At trial, the only issues for the jury to decide were invalidity, willfulness, and damages. F5 asserted that three references anticipated Radware's asserted claims: (1) U.S. Patent No. 6,650,621; (2) the Cisco DistributedDirector ("DD") Device; and (3) the F5 Big-IP/3DNS Device. F5's instant motion covers only the latter two references. Both of Radware's asserted patents are in the same family and claim priority to a parent application that was filed on December 20, 1999. At trial, however, Radware presented evidence that it conceived of the inventions claimed in the '319 and '374 Patents and reduced them to practice before December 20, 1999. The earliest date that Radware's witnesses claimed to have conceived their patented inventions was March or April 1999. Trial Tr. vol. 3, 303:7-9 (Zisapel).

The parties stipulated that certain versions of F5's accused products infringed certain asserted claims. This court's August 22, 2016 order regarding post-trial motions traces much of the relevant factual and procedural history of this case, and this order does not repeat that background information except to the extent necessary to resolve the instant motion. See Dkt. No. 609. --------

The court's instruction to the jury on anticipation read, in relevant part:

A patent claim is invalid if the claimed invention is not new. For the claim to be invalid because it is not new, all of its requirements must have existed in a single device that predates the claimed invention , or must have been described in a single previous publication or patent that predates the claimed invention. In patent law, these previous devices , publications or patents are called "prior art references." If a patent claim is not new we say it is "anticipated" by a prior art reference.
Dkt. No. 552 at 12 (Jury Instr. No. 10) (emphasis added). With respect to invalidity, the jury's verdict form was divided into Part A - Priority, and Part B - Anticipation. Dkt. No. 556 at 2. In Part A, the form asked, for each reference, "Was [reference name] prior art to the asserted patents?" Id. Only if the jury answered "yes" in part A for a reference would they move on to part B, which asked whether F5 had proven by clear and convincing evidence that the asserted prior art reference included all of the requirements of each asserted claim. Id. at 2-3.

On March 15, 2016, the jury found that each of F5's asserted references did not constitute prior art to Radware's asserted patents, and so the jury did not make any factual findings on whether the references included all of the limitations of Radware's asserted claims. Id. at 2-3. On August 22, 2016, this court entered an order resolving several post-trial motions, Dkt. No. 609, and the court entered judgment on September 7, 2016. Dkt. No. 616.

F5 filed the instant motion on October 5, 2016. Dkt. No. 624. Radware filed an opposition on October 19, 2016. Dkt. No. 628. F5 filed a reply on October 26, 2016. Dkt. No. 629.

II. ANALYSIS

Federal Rule of Civil Procedure 59 provides that a new trial may be granted "for any reason for which a new trial has heretofore been granted in an action at law in federal court." Fed. R. Civ. P. 59(a)(1)(a). "Historically recognized grounds" for a new trial "include, but are not limited to, claims 'that the verdict is against the weight of the evidence, that the damages are excessive, or that, for other reasons, the trial was not fair to the party moving.'" Molski v. M.J. Cable, Inc., 481 F.3d 724, 729 (9th Cir. 2007) (quoting Montgomery Ward & Co. v. Duncan, 311 U.S. 243, 251 (1940)). Under Rule 59, "the trial court may grant a new trial, even though the verdict is supported by substantial evidence, if the verdict is contrary to the clear weight of the evidence, or is based upon evidence which is false, or to prevent, in the sound discretion of the trial court, a miscarriage of justice." Wordtech Sys. v. Integrated Networks Solutions, Inc., 609 F.3d 1308, 1313 (Fed. Cir. 2010) (quoting United States v. 4.0 Acres of Land, 175 F.3d 1133, 1139 (9th Cir. 1999)). "[A] district court may not grant or deny a new trial merely because it would have arrived at a different verdict." 4.0 Acres of Land, 175 F.3d at 1139.

"A patent claim is anticipated if a single prior art reference expressly or inherently discloses every limitation of the claim." DDR Holdings, LLC v. Hotels.com, L.P., 773 F.3d 1245, 1252 (Fed. Cir. 2014). "Invalidity by anticipation must be proven by clear and convincing evidence." Id. The parties agree that to show that an asserted device was potentially invaliding "prior art," F5 had to show that the reference was (1) a "single device;" that (2) predated the claimed inventions. Dkt. No. 624 at 3-4; Dkt. No. 628 at 2.

A. "Single Device"

The bulk of the parties' current dispute centers on whether each asserted prior art reference was a "single device" as of the priority date. Both Radware's products embodying the asserted patents and F5's accused products are computer networking devices comprising hardware and software. Either the devices can be installed individually, or groups of the devices can be installed and configured to work together. For example, F5's accused Viprion product comprises a chassis or "rack" in which individual hardware modules can be mounted:

Image materials not available for display.

VIPRION 4480 Chassis TX-281 at 2; see also Trial Tr. vol. 5, 573:10-574:7 (F5 witness Campa describing the accused products).

In explaining the type of system that would constitute a single device meeting the limitations Radware's asserted claims, Radware's technical expert testified that "two modules that may be next to each other or within the same box " could satisfy the language of Radware's claims. Trial Tr. vol. 7, 852:18-19 (Rubin) (emphasis added). Dr. Rubin went on to say modules that are "next to each other connected by a bus or some other very short distance length network" could satisfy Radware's claims. Id. at 853:2-3. Later, on re-direct, Dr. Rubin testified that one could construct a device that satisfies Radware's claims "with multiple modules, multiple devices, components, whatever you call it," as long as the sub-devices are "connected by local area network or by a bus or connected by all of those and being in a single enclosure." Trial Tr. vol. 11, 1668:6-14. Furthermore, Dr. Rubin agreed with F5's counsel that "a series of hardware modules stacked on top of each other encased in a chassis that can be partitioned into different networks infringes the asserted claims." Id. at 1687:8-13.

Claims must be interpreted the same way for invalidity and infringement. Source Search Techs., LLC v. LendingTree, LLC, 588 F.3d 1063, 1075 (Fed. Cir. 2009). F5 asserts that even under Dr. Rubin's interpretation of "device," which F5 argues is too narrow, each of F5's asserted prior art devices would still qualify as a "single device." See Dkt. No. 624 at 8. The court finds that Dr. Rubin's analysis forms an appropriate basis on which to evaluate whether the asserted prior art references constituted "single devices" that were within the scope of the asserted claims.

B. F5 BIG-IP / 3DNS System

F5 argued at trial that when early versions of its BIG-IP and 3DNS units worked together, the resulting "composite device" could practice each of Radware's asserted claims. F5 argues that the clear weight of the evidence shows that F5's BIG-IP and 3DNS were publicly sold and used as a single composite device before Radware's earliest asserted priority date. With respect to the dates, it seems undisputed that the BIG-IP and 3DNS existed in some form prior to Radware's effective filing date of December 20, 1999. For example, F5 introduced user manuals for various early versions of the BIG-IP and 3DNS. See, e.g., TX-660 (BIG-IP Installation & User's Guide ver. 2.0); TX-683 (3DNS ver. 1.04 manual); TX-617 (3DNS ver. 1.03 Installation & User's Guide). Radware even stipulated that these manuals were accessible to the public prior to December 20, 1998. Dkt. No. 419 at 2-3. However, the parties' dispute whether the BIG-IP and 3DNS existed in a "single device" before Radware's priority date.

F5 presented documentation as well as testimony from three witnesses in support of its argument that the BIG-IP and 3DNS were sold together before Radware's effective filing date. For example, a 1998 manual for 3DNS indicates that 3DNS "Integrates seamlessly with a BIG/ip controller network." TX-617 at 21. F5 Software engineer Bryan Skene testified that the 3DNS was intended to work with BIG-IP and that 3DNS was "almost always sold with the BIG-IP." Trial Tr. vol. 8, 1039:23-24, 1040:13-14. He further testified that a 3DNS was usually placed "right next to the BIG-IP." Id. at 1040:15-1041:3. F5's former product manager Jason Needham testified that there were customers using 3DNS and BIG-IP to do ISP load balancing. Trial Tr. vol. 11 at 1536:5-11. F5 senior software architect Peter Thornewell testified that "your 3-DNS and BIG-IP are rack mounted devices, we are supposed to put them in a rack and mount them together." Trial Tr. vol. 9, 1360:16-18.

Radware, on the other hand, argued that the BIG-IP and 3DNS did not become a "single device" until years later when F5 was developing its Link Controller to compete with Radware. Radware points to the following testimony from Skene:

Q. Now do you personally have any of your own experience with respect to combining a BIG-IP 3-DNS device to do load balancing?
A. Yes.
Q. When did you obtain that experience, approximately?
A. Probably in 2001.
Trial Tr. vol. 8, 1041:14-19. Radware also cites the following testimony from Skene:
Q. Do you know if the 3-DNS and BIG-IP were ever combined as a single device?
A. Yes.
Q. Do you know about when that was?
A. I think it was around the beginning of 2002.
Id. at 1041:4-8. Radware also pointed to an email exchange among F5 employees from around 2000 indicating that they were impressed with the ISP load balancing functionality of Radware's products. See TX-64 at 9 ("I just read the PDF on the linkproof device and it sits behined [sic] the access routers to the ISP's. How does a device which sits behind an access router influence how traffic on the opposite side gets to those access routers.").

After considering the proffered testimony and exhibits, the court is not persuaded that a new trial is justified with respect to the jury's finding that the BIG-IP / 3DNS system was not "prior art." The evidence that the BIG-IP and 3DNS were designed to work together in the 1998 timeframe went unrebutted, but evidence of compatibility does not compel a conclusion that the two were a "single device." Moreover, Skene's testimony that a 3DNS was usually placed "right next to the BIG-IP," Trial Tr. vol. 8, 1040:15-1041:3, was not specific to a particular time. The jury could have reasonably inferred from Skene's testimony and from the cited exhibits that BIG-IP and 3DNS did not become a single device until 2001 or 2002, after the priority date of Radware's asserted patents. Accordingly, F5's motion for a new trial as to invalidity based on the BIG-IP / 3DNS is denied.

C. Cisco DistributedDirector

F5 also argues that the clear weight of the evidence shows that the Cisco DD was publicly sold and used as a single device before Radware's earliest asserted priority date. As with the BIG-IP / 3DNS, the court agrees with F5 with respect to the date that the Cisco DD was available. Radware stipulated that the Cisco Distributed Director version 11.1.20 IA was compiled July 2, 1998, and the associated Cisco 2500 Series Router IOS version 12.02 software was compiled December 3, 1998. Trial Tr. vol. 9, 1337:4-11. Moreover, F5 presented unrebutted testimony that the Cisco DD was publicly available in some form in 1998, before Radware's earliest asserted priority date of April 1998. Trial Tr. vol. 9, 1238:6-8 (Darwin); id. at 1303:14-1304:1, 1315:22-1316:8 (Delgadillo testimony regarding Cisco software release dates). Once again, however, the parties' dispute whether the Cisco DD existed as a "single device" as contemplated by Radware's claims before Radware's priority date.

In addition to Cisco documentation from the late 1990s, e.g., TX-657, F5 principally relies on the testimony of two former Cisco employees to argue that the Cisco DD software, in conjunction with software on Cisco routers, could perform ISP load balancing before Radware's December 1999 priority date. Dean Darwin, a former global account manager at Cisco, testified that in 1998 or 1999, Cisco was selling the Distributed Director and that it could perform ISP load balancing. Trial Tr. vol. 9, 1239:12-16. He testified that Microsoft used the Cisco DD "as part of a solution that we were selling to them that would keep the links open" at sites in remote areas in which ISP connectivity was costly, and availability was subpar. Id. at 1239:24-1240:7. Kevin Delgadillo, a former Cisco Product manager who started working for F5 during the pendency of this lawsuit, testified as follows regarding the Cisco DD software:

Q. And how about the software itself, how familiar with that software of distributed director were you as a result of your role as product manager?

A. I was very familiar with that because there were two components to the solution. There was the IOS software that ran on a routing platform for the Distributed Director itself , which was not really a router anymore, it was more of an intelligent DNS server. And then there was a different component that would be run in the data center near the web servers called a DRP agent , which again was IOS
software running on the routing platform.
Id. at 1298:19-1299:1 (emphasis added). Delgadillo testified that the Cisco DD acted as a DNS server and worked with DRP agents running on Cisco routers to measure round trip time to destinations and select destination servers accordingly. See id. at 1302:16-1303:13.

Radware seizes on Delgadillo's testimony that the Cisco DD composite device had "two components." Radware argues that the Cisco DD and the DRP agents running on Cisco 2500 routers were actually separate devices. Radware points out that unlike in the case of the BIG-IP and 3DNS, F5 cites no evidence indicating that the Cisco DD and the Cisco 2500 routers were sold in combination. Radware also argues that the independent version numbers and compilation dates for the Cisco DD and Cisco 2500 Series Router IOS mentioned above suggest that the Cisco DD and router software were not part of a single device.

As in the case of the BIG-IP / 3DNS system, the court is not persuaded that a new trial is justified with respect to the jury's finding that the Cisco DD composite system was not "prior art." While Radware's arguments do not compel a finding that the Cisco DD was not bundled with Cisco 2500 routers in a single device, F5 had the burden of proving by clear and convincing evidence that Cisco did make such a device. The jury could have reasonably inferred from F5's inability to cite any instances of the Cisco DD and Cisco 2500 routers being sold together that the products were not actually bundled together as a single device in 1998. The court is not firmly convinced that the jury made a mistake in concluding that F5 failed to meet its burden. Accordingly, F5's motion for a new trial as to invalidity based on the Cisco DD is denied.

III. ORDER

For the foregoing reasons, the court DENIES F5's motion for a new trial.

IT IS SO ORDERED. Dated: October 31, 2016

/s/_________

Ronald M. Whyte

United States District Judge


Summaries of

Radware, Ltd. v. F5 Networks, Inc.

UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA SAN JOSE DIVISION
Oct 31, 2016
Case No. 5:13-cv-02024-RMW (N.D. Cal. Oct. 31, 2016)
Case details for

Radware, Ltd. v. F5 Networks, Inc.

Case Details

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: Oct 31, 2016

Citations

Case No. 5:13-cv-02024-RMW (N.D. Cal. Oct. 31, 2016)