• HQ H284032
• Apr 07, 2017

• Type : 19 U.S.C. § 1337; Unfair Competition • HTSUS :
•  Revokes:   H277148     Related:   H277148; H259071; H067500; H121519; H242025; H242026; H226615   

OT:RR:BSTC:IPR H284032 DAX/JW

Mr. Bert C. Reiser Latham & Watkins LLP
555 Eleventh Street, N.W., Suite 1000
Washington, D.C. 20004-1304

RE: Ruling Request; U.S. International Trade Commission; Limited Exclusion Order; Investigation No. 337-TA-944; Certain Network Devices, Related Software and Components Thereof

Dear Mr. Reiser:

This ruling letter, issued under 19 C.F.R. § 177, is the result of an inter partes proceeding that the Intellectual Property Rights Branch, Regulations and Rulings, U.S. Customs and Border Protection (“CBP”), administered pursuant to section 337 of the Tariff Act of 1930, as amended, 19 U.S.C. § 1337, and that involved your client, Arista Networks, Inc. (“Arista”), and Cisco Systems, Inc. (“Cisco”) as the two parties with a direct and demonstrable interest in the question presented by their respective ruling requests. The question before CBP is whether certain Arista network devices (switches and routers) running the re-designed version 4.16.6M of its Extensible Operating System (“the EOS Redesign”) are subject to the exclusion order issued by the U.S. International Trade Commission (“Commission” or “ITC”) in the above-referenced investigation.

I. FACTS:

ITC Investigation No. 337-TA-944

The Commission instituted Investigation No. 337-TA-944 on January 27, 2015, based on a complaint filed on behalf of Cisco. See 80 Fed. Reg. 4314-15 (Jan. 27, 2015). Cisco alleged violations of section 337 of the Tariff Act of 1930 (19 U.S.C. § 1337) by reason of infringement of various claims of U.S. Patent No. 7,162,537 (“the `537 patent”); U.S. Patent No. 8,356,296 (“the `296 patent”); U.S. Patent No. 7,290,164 (“the `164 patent”); U.S. Patent No. 7,340,597 (“the `597 patent”); U.S. Patent No. 6,741,592 (“the `592 patent”); and U.S. Patent No. 7,200,145 (“the `145 patent”).

On February 2, 2016, the presiding Administrative Law Judge (“ALJ”) issued his final initial determination (“ALJ FID”), finding that the '537, '592 and '145 patents were infringed by Arista’s accused products. See ALJ ID at 292. On April 11, 2016, the 

Commission determined to review the ALJ FID in-part. See 81 Fed. Reg. 42375-76 (June 29, 2016). On June 23, 2016, the Commission affirmed the ALJ ID in-part and issued a limited exclusion order and a cease and desist order against Arista. Specifically, the former directed that:

[n]etwork devices, related software and components thereof that infringe one or more of 1, 2, 8-11, and 17-19 of the '537 patent; claims 6, 7, 20, and 21 of the '592 patent; and claims 5, 7, 45, and 46 of the '145 patent that are manufactured abroad by or on behalf of, or imported by or on behalf of, Respondent, or its affiliated companies, parents, subsidiaries, licensees, or other related business entities, or its successors or assigns, are excluded from entry for consumption into the United States, entry for consumption from a foreign trade zone, or withdrawal from a warehouse for consumption, for the remaining terms of the patents, except under license of the patent owner or as provided by law, and except for service, repair, or replacement articles imported for use in servicing, repairing, or replacing network devices under warranty or service contracts, for identical articles, that existed as of the date of this Order.

See Investigation No. 337-TA-944 Limited Exclusion Order (“944 LEO”) (June 23, 2016).

On September 28, 2016, the Commission instituted a formal enforcement proceeding based on a complaint filed on behalf of Cisco on August 26, 2016. See Investigation No. 337-TA-944 Notice of Institution of Formal Enforcement Proceeding. Cisco’s enforcement complaint includes an allegation that “[t]esting of allegedly redesigned products . . . demonstrates that Arista’s products continue to include Cisco’s patented SysDB technology, and continue to infringe the U.S. Patent No. 7,162,537[.]” Cisco Complaint (dated August 26, 2016, at pg. 5, ¶ 1.3). Cisco did not allege continued infringement of the ‘592 or ‘145 patents. Id.

B. CBP HQ Ruling H277148

On June 27, 2016, pursuant to 19 C.F.R. § 177, Arista submitted a request to CBP for an administrative ruling whether certain Arista network devices running the re- designed version 4.16.6M of its Extensible Operating System (“EOS Redesign”) were subject to the 944 LEO. See Arista Ruling Request (dated June, 27, 2016). On November 18, 2016, CBP issued HQ Ruling H277148 holding:

[t]he Arista Redesign EOS at issue is not covered by claims 1, 2, 8-11, and 17-19 of the '537 patent; claims 6, 7, 20, and 21 of the '592 patent; or claims 5, 7, 45, and 46 of the '145 patent. Therefore, the switches in question are not within the scope of the

2 

ITC’s limited exclusion order in Investigation No. 337-TA-944, and may be entered for consumption into the United States.

See CBP HQ Ruling H277148 (November 18, 2016).

Subsequently, on December 14, 2016, Cisco submitted to CBP a ruling revocation request pursuant to 19 C.F.R. § 177, requesting that CBP revoke HQ Ruling H277148. See Cisco Revocation Request (dated December 14, 2016). Through this submission, Cisco alleged that Arista’s switches running the EOS Redesign still infringed the ‘537 patent. Id. at 2. No contentions were raised with respect to the ‘592 patent or the ‘145 patent. Id. On January 13, 2017, under 19 U.S.C. § 1625(c)(1), as implemented by 19 C.F.R. § 177.12(b), CBP revoked HQ Ruling H277148 (“the revoked ruling”). See CBP Letter to B. Reiser (dated January 13, 2017).

C. The ‘537 Patent

The subject matter of the `537 patent, as indicated in the abstract, pertains to a method and system in a router device for externally managing router configuration data in conjunction with a centralized database subsystem, wherein the centralized database provides external management registration and unregistration for various managing router subsystems associated with that database system.




Figure 2 above shows a diagram of an internetwork operating system in accordance with the ‘537 patent. In this figure, examples of subsystem applications appear in the various blocks, with the centralized database subsystem appearing as “sysDB”1 in block 26 and the centralized database as the “sysDB TREE” in block 42. During the investigation at the ITC, the ALJ ID referred to the centralized database of the legacy products as “Sysdb” in addition to identifying sysDB in the ‘537 patent as

1 As in the revoked rulings, CBP will use “sysDB” when referring to the ‘537 patent and “Sysdb” when referring to a feature of the Arista switches.

3  both a “centralized database” and a “database subsystem” while the Commission Opinion interpreted both Sysdb and sysDB as referencing the “same thing,” namely a centralized database. See ALJ ID at 6, 9, 14; Comm’n Op. at 8, FN 10. While the ‘537 patent claims refer to “database” and “database subsystem” as separate terms in the claim limitations, the relationship between the two can be expressed the way in which both occupy the same structure as shown in block 26 of Figure 4, as depicted below.




This example of external management provides modularity and independence of the various subsystems of the router device. See `537 patent at Col. 4, ln. 26-29. Router devices, as a general matter, are responsible for carrying information packets between a network of coupled devices. See `537 patent at Col. 1, ln. 19-22. To control this task, a user can employ a configuration subsystem to configure the routing device’s operations. Id. at ln. 22-25. The configuration subsystem determines the affected client subsystem based on the information in the user’s configuration command. Id. at ln. 30- 35. In the prior art, however, the client subsystems often had multiple dependencies between each other requiring client subsystems to consistently access each other’s information before having the necessary data to propagate a user’s input task with responsibilities shared among several subsystems. Id. at ln. 37-47. Compounding this issue was the difficulty encountered when either a central database retained all the needed information or when all the data was kept outside a central database, leading to heavier burdens on the central database in the former and non-modular external components in the latter. Id. at Col. 2, ln. 58-64; Col. 3, ln. 7-12. As such, the ‘537 patent claimed a method and system for external data management by client subsystems of router configuration data stored in a centralized database system. Specifically, subsystems can send a registration request to sysDB to externally manage that configuration data. Id. at Col. 5, ln. 18-25. In doing so, the request indicates what router configuration data is requested from what part (or “tuple”) of the sysDB tree; if the externally managed data is to be stored in the sysDB and, if so, when the data in the sysDB would expire; and the address where sysDB can get the externally managed data. Id. at ln. 18-32. In response to the request, sysDB provides whatever non- expired data exists in its cache to the address that the requesting subsystem specified and sets up the tuple to be externally managed. Id. at ln. 32-40. This exchange of information is illustrated below in Figure 5 and occurs through interaction between the sysDB’s external managing unit and the requesting subsystem’s local managing unit that were both depicted above in Figure 4. Id. at ln. 8-11.

4 


D. Claims at Issue

With respect to the ‘537 patent, the 944 LEO excludes Arista’s “[n]etwork devices, related software and components thereof that infringe one or more of 1, 2, 8- 11, and 17-19. . . .” See 944 LEO. Of these claims, claims 1, 10, and 19 are independent claims. See ‘537 patent. Independent claims 1, 10, and 19 of the '537 patent are listed below with the limitations relevant to this ruling letter bolded and underlined.

Claim 1 1. A method for reducing computational overhead in a centralized database system by externally managing router data in conjunction with a centralized database subsystem, said database subsystem operatively coupled for communication with a plurality of router subsystems one of which is a first managing subsystem, comprising:

a) transmitting a management registration request by said first managing subsystem to said database subsystem, said registration request indicating router configuration data for which said first managing subsystem is requesting to provide external management services, said router configuration data managed by said database system and derived from configuration commands supplied by a user and executed by a router configuration subsystem before being stored in said database;

5  b) receiving said management registration request by said database subsystem; and

c) registering said first managing subsystem for external management by said database subsystem.

Claim 10

10. A program storage device readable by a machine, tangibly embodying a program of instructions executable by the machine to perform a method for reducing computational overhead in a centralized database system by externally managing router data in conjunction with a centralized database subsystem, said database subsystem operatively coupled for communication with a plurality of router subsystems one of which is a first managing subsystem, said method comprising:

transmitting a management registration request by said first managing subsystem to said database subsystem, said registration request indicating router configuration data for which said first managing subsystem is requesting to provide external management services, said router configuration data managed by said database system and derived from configuration commands supplied by a user and executed by a router configuration subsystem before being stored in said database;

receiving said management registration request by said database subsystem; and

registering said first managing subsystem for external management by said managing subsystem.

Claim 19 19. In a router device having a processor and memory, a router operating system executing within said memory comprising:

a database subsystem;

a plurality of client subsystems, each operatively coupled for communication to said database subsystem, one of said client subsystems configured as a managing subsystem to externally manage router data upon issuing a management request to said database subsystem; and

(c) a database operatively coupled to said database subsystem, said database configured to store router configuration data and delegate management of router configuration data to a management subsystem that requests to manage router configuration data, said router configuration data managed by said database system and derived from configuration commands supplied by a user and executed by a router configuration subsystem before being stored in said database.

E. Claim Constructions from the Commission for the ‘537 Patent

6 

Claims 1, 10, and 19 have many claim limitations in common and differ only in the way that the invention has been claimed (namely, and respectively, a method claim, machine-readable instructions, and an apparatus claim). See ALJ ID at 65. As such, the ALJ's claim construction of disputed claim terms grouped the analogous claim limitations of each claim and applied the same construction to each grouping of analogous claim limitations. These claim constructions are provided below :

CLAIM TERM CONSTRUCTION  “externally managing router data” (claims 1 and 10); “externally manage router data” (claim 19); “external management” (claims 1 and 10); “management of” (claim 19) The claim term “external management” and variations thereof do not need construction. ALJ ID at 55-56. The meaning of “external management” is plain to a person having ordinary skill.” Id. at 55. “A person having ordinary skill would understand ‘external management’ to mean that the subsystem, which is external to the centralized database system, manages the data.” Id.  “management registration request” (claims 1 and 10); “management request” (claim 19) “[A] request to register to provide external management services.” Id. at 57.  “router configuration data” (claims 1, 2, 10, 11, and 19) “[T]he claim term ‘router configuration data’ does not need construction.” Id. at 58-59.  “said database” (claims 1 and 10) “[T]his claim term is not indefinite.” Id. at 60.  “reducing computational overhead in a centralized database system” (claims 1 and 10) “[T]his claim term is limiting.” Id. at 61-63. “It is further determined that the claim term ‘reducing computational overhead in a centralized database system’ is construed to mean ‘reducing the amount of computation in a centralized database system.’” Id. at 63.  “said router configuration data managed by said database system and derived from configuration commands supplied by a user and executed by a router configuration subsystem before being stored in said database” (claims 1, 10, and 19) “[T]he proper construction of this term requires the storage of router configuration data.” Commission Opinion at 8-10.  

7  F. Arista’s Legacy Products

In the products found to infringe, the legacy EOS was a router operating system on router devices with a processor and memory in which the legacy EOS ran and where the legacy EOS processes were further found to “exchange state through an in-memory database.” See ALJ ID at 65. The ALJ found these features to satisfy the “[i]n a router device…within said memory comprising:” limitation of claim 19. Id.

Additionally, in the legacy products, the Sysdb interacted with its agents through “mounting” akin to the patent’s teaching of the “database subsystem” where the centralized database subsystem receives the management registration request from a router or client subsystem configured as a managing subsystem that registers to provide external management services. Id. at 65-66. In this manner, Sysdb satisfied the claim limitation of “a database subsystem” in claim 19. Id. at 65.

Further, as each agent in the legacy product generally handled a particular feature or set of related features (e.g., the MLAG agent manages MLAG data) and was coupled to EOS’s Sysdb for such handling, the ALJ found the EOS agents to satisfy the “plurality of client subsystems, each operatively coupled for communication to said database subsystem” limitation of claim 19. Id. at 66. With respect to that coupling, the complainant’s expert, upon whom the ALJ relied, described the agents as meeting that coupling limitation since it was “the EOS agents that connect to Sysdb.” ITC Complainant’s Exhibit 0007C at Q/A 128.

[ ] See ALJ ID at 66, FN 23. [

] Id. [ ] Id. The ALJ found the EOS agents externally managed data by [ ] the data in Sysdb, since an agent [ ] Id. In the legacy product, [

] Id. at 67-68. In this manner, the legacy product’s [

] Id. at 69. As such, the agents in the legacy products practiced claim 19’s limitation of “one of said client subsystems configured as a managing subsystem to externally manage router data.” Id. at 66.

The management request, or [ ], that the agent sent to Sysdb indicated the [

8  ] Id. at 69-70. The ALJ found that functionality to satisfy “upon issuing a management request to said database subsystem” limitation of claim 19.2 Id. at 70. The legacy product’s Sysdb included a [

]; further, the [ ] Id. This arrangement satisfied the “database operatively coupled to said database subsystem, said database configured to store router configuration data” in claim 19. Id. In the legacy products, Sysdb processed write-mount requests and allowed the requesting agent to externally manage data, specifically router configuration data described further below. Id. at 70-71. The ALJ found this functionality to satisfy claim 19’s limitation requiring the database’s configuration to “delegate management of router configuration data to a management subsystem that requests to manage router configuration data.” Id.

An overview of the processes described above in the legacy products is provided in the figure below. Arista Supplement (dated September 19, 2016 at 2). As depicted in the figure below, a process called Proc Manager (“ProcMgr”) would start an agent (here, the Fhrp agent) and the agent could then send write mount requests to Sysdb. Id. The write mount request is indicated as [ ] Id.

[ ]

With respect to router configuration information, the legacy products [

] Id. at 71. The [

] Id. at 71-72. As such, the [

_____________________________________________ 2 [

] See ITC Complainant’s Exhibit 0007C at Q/A 129.

9

] Id. The

] Id. at 72. The ALJ found that this [

] met the claim limitation for “router configuration data managed by said database system and derived from configuration commands supplied by a user and executed by a router configuration subsystem before being stored in said database.” Id. at 71-72.

G. Arista’s EOS Redesign

Arista purports to have removed the [ ] from the agents to Sysdb found to infringe the ‘537 patent. Arista Ruling Request at 4. To effect this change, Arista claimed that the agents will no longer send to Sysdb [ ] that specify the details of what the agent wishes to mount. Id. at 13, 16. Instead, the EOS Redesign will start agents with every [ ] the agent could need in place (except for the CLI agent which will retain its [ ]), as opposed to the ad-hoc request-based method used in the legacy products. Id. at 16, 17. The EOS Redesign achieves this end with a program called “ProcMgr” (short for Process Manager) which starts and stops agents. Id. at 17. Specifically, “[t]he ProcMgr calls the [ ] for each agent that it is starting that needs a connection to Sysdb established before it starts.” Cisco Submission (dated February 22, 2017 Exhibit A at 22 (citing Sweeney 944E Dep. Tr. At 136:24-137:5)). ProcMgr starts and stops EOS agents, but cannot mount state from Sysdb, read data from Sysdb, or write data to Sysdb. Arista Supplement I at 2, 4. Further, ProcMgr does not use the AttrLog communications protocol that all EOS agents use to communicate with Sysdb, nor does ProcMgr have a [ ] profile, a [ ] of its own, or access to the [ ] profile or [ ] table of any EOS agent. Id. at 4.

Once ProcMgr begins the process to start an agent, ProcMgr issues a [ ] command, through the [ ], which includes the agent’s name to another process, [ ] which then creates a connection to Sysdb and passes information from the [ ] command to Sysdb. Arista Ruling Request at 17; Arista Submission dated February 2, 2017 at 36. As a side note, to communicate with Sysdb – including sending [ ], or any other communications – an agent must first be connected to Sysdb by a socket. Arista Supplement I at 2. After the connection is made, [ ] sends data about the connection back to ProcMgr. Arista Ruling Request at 17. The [ ] command’s full message forwarded to Sysdb reads as, [

] Id. at 18. As such, Arista claimed that only the name of the agent was passed from [ ] to Sysdb. Id. According to Arista, the [ ] command is always sent before the agent has been started. Id. In other words, when the [ ] command is sent, the agent is

10 

[ ]. See e.g., Arista Submission (dated February 23, 2017) at 7-12. The representative data processing in the EOS Redesign is pictured below. Arista Ruling Request at 17.

[ ]3

Next, for each of the over eighty EOS agents, Arista claims to have created a profile with every [ ] the agent may need or want; the profile identifies those potential mounts to Sysdb. Id. at 19. Once Sysdb receives the [ ] command with the agent’s name, Sysdb reads the profile and populates Sysdb’s [ ] table copy with the mounts. Id. At that point, Sysdb writes the table into the socket [ ] made in the prior step, along with a copy of all the state that was mounted. Id. at 19-20. As the agent usually is not started yet, the agent does not read the connection’s other end. Id. at 19. Linux then puts the data Sysdb sent (including the [ ] table, or a portion thereof) into a kernel socket buffer, where the data remains until the agent exists and can read the data at the connection’s end. Id. Linux is an open source operating system based in part on the Unix operating system. Id. at FN 9. These steps can be illustrated as follows:

[ ]

Once the Sysdb connection is made, the agent can be started and pre-loaded with the [ ] table. Id. at 20. An overview of the connection establishment process in the redesign is provided in the picture below:

__________________________ 3 Arista appears to use the phrase [

]. See Cisco Post-Proceeding Submission (dated February 23, 2017) at 14; see also Arista Post-Proceeding Submission (dated February 23, 2017) at 8-10. 11 

[ ] See ANl-ITC-944E-00000002 (Figure 2).

Next, as in the legacy product, ProcMgr processes the code to start the agent using the standard Linux socket mechanism to send data from Sysdb to the agent. Arista Ruling Request at 20. Before the agent starts, the I ] table data will remain in the kernel buffer associated with the socket connection between Sysdb and the agent and then get retrieved by the agent with standard Linux system calls when the agent starts. Id. These calls will read any data from the kernel and are begun once the agent is apprised there is some kind of data in the kernel awaiting the agent. Id. at 20-21. Sysdb does not act in response to such calls or get notice of such calls. These steps can be seen illustrated in the following:

[ ]

As to the process of reading data from the kernel buffer, the agent causes the Linux kernel to copy data from the kernel buffer to the process's own buffer when the agent reads data from a socket. Arista Supplement II at 2. The agent reads this data out of the socket through a "read system call." Id. Generally, read system calls have only three parameters: the identity of the socket that the process will read, the address of the process buffer to which the data will be copied, and the maximum data that can be retrieved. Id. Once the agent makes the read system call, the Linux kernel checks for data in the buffer associated with the designated socket connection and copies the data from the kernel buffer to the process's buffer if so, and indicates there is no data if not. Id. Upon copying the data to the process's buffer, the kernel buffer will also

12 

indicate the number of bytes copied, after which the data will no longer be available in the kernel buffer. Id. However, the kernel cannot receive that system call until after Sysdb has made a complete pass over the [ ] profile, updating its own internal records to reflect that Sysdb has established the mounts, before Sysdb writes any of the state (including the [ ] table) to the socket and so before the [ ] table or any other state is placed in the kernel buffer. Id. at 5. In other words, the agent will not issue a read system call to the kernel without all mounts being in place and Sysdb initiating writing [ ] tables to the socket. Id. Relatedly, there is no address (or function) that Sysdb may use to get router data, as Sysdb never requested data copies from agents. Arista Supplement III at 15.

As to the CLI agent, the EOS Redesign will only allow the [ ] or [ ] flags on mount requests to read as true for the CLI agent, meaning the mount request will be for a [ ]; [ ] would set [ ] to false and allow the database’s state to change. Arista Ruling Request at 21. Arista also programmed the EOS Redesign to have Sysdb shut down the socket connection should an agent be modified to permit the agent to send a [ ] to Sysdb. Id. at 22.

ISSUE:

Whether Arista switches running the EOS Redesign are covered by the asserted claims of the `537, `592, and `145 patents and therefore fall within the scope of the Commission’s limited exclusion order from Investigation No. 337-TA-944, such that they would be excluded from entry for consumption into the United States.

LAW AND ANALYSIS:

A. Ability to Issue Ruling Letter Under 19 C.F.R. § 177

A preliminary question is whether, in light of either Arista’s or Cisco’s separate appeal of the Commission’s Final Determination from the underlying investigation, as discussed in further detail below, preclude CBP from ruling on the question presented, specifically under 19 C.F.R. § 177.7 (which addresses situations in which no CBP ruling will be issued).

As recognized in the revoked ruling, it has long been the position of U.S. Customs and Border Protection (including its predecessor, the U.S. Customs Service) and the Treasury Department that, to further the sound administration of the Customs and related laws, “persons engaging in any transaction affected by those laws fully understand the consequences of that transaction prior to its consummation[.]” For this reason, CBP “will give full and careful consideration to written requests from importers and other interested parties for rulings or information setting forth, with respect to a specifically described transaction, a definitive interpretation of applicable law, or other appropriate information.” 19 C.F.R. § 177.1(a); see also Proposed Revision Relating to Issuance of Administrative Rulings by Headquarters Office, Notice of Proposed

13 

Rulemaking, 40 Fed. Reg. 2437 (January 13, 1975) and Issuance of Administrative Rulings by Headquarters Office, Final Rule, 40 Fed. Reg. 31929 (July 30, 1975) (promulgating this provision of 19 C.F.R. § 177 as it appears, unchanged, in the current edition of the Customs regulations).

“Generally, a ruling may be requested under the provisions of this part only with respect to prospective transactions—that is, transactions which are not already pending before a Customs Service office by reason of arrival, entry, or otherwise.” Id. Additionally, “[n]o ruling letter will be issued with regard to transactions or questions which are essentially hypothetical in nature or in any instance in which it appears contrary to the sound administration of the Customs and related laws to do so.” 19 C.F.R. § 177.7(a). Moreover, “[n]o ruling letter will be issued with respect to any issue which is pending before the United States Court of International Trade, the United States Court of Appeals for the Federal Circuit, or any court of appeal therefrom.” 19 C.F.R. § 177.7(b) (emphasis added).

Notably, on September 23, 2016, Arista filed its opening brief with the U.S. Court of Appeals for the Federal Circuit (“Federal Circuit”) under 19 U.S.C. § 1337(c) and 28 U.S.C. § 1295(a)(6), appealing, inter alia, the Commission’s Final Determination (and related Commission Opinion) of June 23, 2016, which resulted in the limited exclusion order at issue directing the exclusion from entry of Arista switches covered by certain patent claims. See Arista Networks, Inc v. U.S. International Trade Commission, Court No. 16-2563. Importantly for purposes of this ruling letter, the appeal to the Federal Circuit involves the Commission’s Final Determination, resulting from the investigation under section 337, that a violation of this section exists based on the legacy products accused and found to infringe the relevant patent claims included in the limited exclusion order. The Commission’s Final Determination whether there is a violation of section 337 by those legacy products involves an issue that is separate and distinct from the administration of the exclusion order by Customs4 and its authority, under 19 C.F.R. §§ 177 or 174, to determine whether an unadjudicated article may be entered into the United States in light of an exclusion order (particularly where the article in question is a new or modified product or one that allegedly is not subject to exclusion from entry for some other reason or development, such as a license or patent exhaustion defense).

CBP has repeatedly adhered to the position that appealing a Commission Final Determination to the Federal Circuit does not foreclose CBP from ruling on the admissibility of other products that were not adjudicated during the investigation at the Commission. See CBP HQ Ruling H259071 at 8 (dated December 10, 2014) (“CBP has issued ruling letters during the pendency of appeals at the Federal Circuit that involve a final determination of the Commission because, although the general matter was on appeal, the particular issue before the Federal Circuit [involving the legacy ___________________________ 4 Notwithstanding the above, a Commission Final Determination, of course, is binding as to those legacy products found to infringe and whose compulsory exclusion, if the legacy products attempted entry, would not be a protestable decision under 19 U.S.C. § 1514(a)(4) and, as such, would not require or receive an administrative ruling or protest decision from Customs. 14 

products found to infringe] was not the subject of the administrative ruling.”) (emphasis added); see also CBP HQ Ruling H067500 (dated August 5, 2009).

Since, as noted above, Arista’s appeal of the Commission’s Final Determination does not involve network devices running the EOS Redesign, the subject of this ruling request is not an issue pending before the Federal Circuit and, consequently, CBP is not precluded under 19 C.F.R. § 177.7(b) from ruling on the admissibility of the redesigned switches, just as it would not be precluded from issuing a protest decision under 19 C.F.R. § 174 through an application for further review challenging an exclusion from entry of an article that was not adjudicated by the Commission during the underlying investigation but may be subject to an exclusion order issued by the Commision.

B. Treatment of Confidential Business Information Submitted Under 19 C.F.R. § 177

As for another preliminary matter, Arista has requested confidential treatment in connection with this ruling request for certain information from its submissions, claiming that the disclosure of such information would likely cause substantial harm to its competitive position and, on this basis, seeks to have the information in question redacted from any ruling that is published in accordance with 19 U.S.C. § 1625(a). Disclosure of information related to administrative rulings under 19 C.F.R. § 177 is governed by 31 C.F.R. § 1, 19 C.F.R. § 103, and 19 C.F.R. § 177.8(a)(3). See 19 C.F.R. § 177.10(a). Moreover, the determination whether to include or redact information within a published ruling is guided by various federal laws—and the relevant standards for their application—that involve confidentiality and disclosure, to include the Freedom of Information Act (“FOIA”) (5 U.S.C. § 552), the Trade Secrets Act (18 U.S.C. § 1905), and the Privacy Act of 1974 (5 U.S.C. § 552a). See CBP HQ Ruling H121519 at 1 (dated February 8, 2011).

Congress enacted FOIA to overhaul the earlier public-disclosure section of the Administrative Procedure Act that gradually became more “a withholding statute than a disclosure statute.” Milner v. Dep't of the Navy, 562 U.S. 562, 565 (quoting EPA v. Mink, 410 U.S. 73, 79 (1973). Accordingly, there is a strong presumption in favor of disclosure, which is consistent with the purpose as well as the plain language of the Act. United States Dep't of State v. Ray, 502 U.S. 164, 173 (1991). Thus, FOIA mandates that an agency disclose certain information unless it falls within one of nine exemptions. Milner, at 565. These exemptions are “explicitly made exclusive,” EPA, at 79, and must be “narrowly construed,” FBI v. Abramson, 456 U.S. 615, 630 (1982).

Exemption 4 of FOIA, which is especially relevant here, provides that FOIA does not apply to matters that are “trade secrets and commercial or financial information obtained from a person and privileged or confidential.” 5 U.S.C. § 552 (b)(4); see also Chrysler Corp. v. Brown, 441 U.S. 281, 291 (1979). Furthermore, it is worth noting that, as a general matter, the proscriptions of the Trade Secrets Act are “at least co- extensive with Exemption 4 of FOIA . . . [such that], unless another statute or a

15 

regulation authorizes disclosure of the information, the Trade Secrets Act requires each agency to withhold any information it may withhold under Exemption 4 of the FOIA.” Venetian Casino Resort, LLC v. EEOC, 530 F.3d 925, 932 (D.C. Cir. 2008) (quoting Canadian Comm. Corp. v. Air Force, 514 F.3d 37, 39 (D.C. Cir. 2008) (emphasis added); see also Dow Chem. Co. v. United States, 476 U.S. 227, 234, n. 2 (1986) (“Dow's fear that EPA might disclose trade secrets revealed in these photographs appears adequately addressed by federal law prohibiting such disclosure generally under the Trade Secrets Act, 18 U.S.C. § 1905, and the Freedom of Information Act, 5 U.S.C. § 552(b)(4).”).

That said, “Congress did not design the Freedom of Information Act exemptions to be mandatory bars to disclosure.” Chrysler Corp., 441 U.S. at 293, 294 (“We therefore conclude that Congress did not limit an agency's discretion to disclose information when it enacted the FOIA.”) (emphasis added); see also GTE Sylvania v. Consumers Union of United States, 445 U.S. 375, 378, n. 2 (1980) (“The theory of the so-called ‘reverse Freedom of Information Act’ suit, that the exemptions to the Act were mandatory bars to disclosure and that therefore submitters of information could sue an agency under the Act in order to enjoin release of material, was squarely rejected in Chrysler Corp.”).

The test, for administrative rulings under 19 C.F.R § 177, to determine whether certain information is confidential (and therefore properly redacted from any published ruling) is identical to the federal government-wide standard for disclosure under FOIA Exemption 4 as it relates to the likelihood of substantial harm to the competitive position of the person submitting the information and requesting that it be withheld from publication. As with FOIA, the basic objective of administrative rulings under 19 C.F.R. § 177 favors disclosure to provide notice to interested persons of the reasons for the agency’s position and its decision-making process. See Chrysler Corp., 441 U.S. at 290, n. 10 (“We observed in Department of Air Force v. Rose that ‘disclosure, not secrecy, is the dominant objective of the Act.’ The legislative history is replete with references to Congress' desire to loosen the agency's grip on the data underlying governmental decisionmaking.”) (internal citation omitted). Notably, under 19 C.F.R. § 177.8(a)(3) referenced above, there is a general presumption that “[n]o part of the ruling letter, including names, addresses, or information relating to the business transactions of private parties, shall be deemed to constitute privileged or confidential commercial or financial information or trade secrets exempt from disclosure pursuant to the Freedom of Information Act, as amended (5 U.S.C. 552), unless, as provided in §?177.2(b)(7), the information claimed to be exempt from disclosure is clearly identified and the reasons for the exemption are set forth.”

In turn, 19 C.F.R. § 177.2(b)(7) provides that “[i]nformation which is claimed to constitute trade secrets or privileged or confidential commercial or financial information regarding the business transactions of private parties the disclosure of which would cause substantial harm to the competitive position of the person making the request (or of another interested party), must be identified clearly and the reasons such information should not be disclosed, including, where applicable, the reasons the disclosure of the

16 

information would prejudice the competitive position of the person making the request (or of another interested party) must be set forth.”

Significantly, § 177.2(b)(7) was added to this section through rulemaking (T.D. 75-186, 40 Fed. Reg. 31929, July 30, 1975) that promulgated a final rule based on a previous notice of proposed rulemaking (40 Fed. Reg. 2437, January 13, 1975) where the text as proposed did not contain (b)(7) or any other provision addressing the treatment of putative confidential business information. The “substantial harm to a competitive position” standard adopted in § 177.2(b)(7) to handle confidentiality issues is identical to that standard for FOIA Exemption (b)(4) contemporaneously established by the U.S. Court of Appeals for the District of Columbia Circuit (“D.C. Circuit”) in National Parks & Conservation Asso. v. Morton, 498 F.2d 765 (D.C. Cir. 1974). See N.H. Right to Life v. HHS, 136 S. Ct. 383; 193 L. Ed. 2d 412; 2015 U.S. LEXIS 7169 (2015) (Thomas, J., dissenting from denial of the petition for writ of certiorari). In 1974, a year before the promulgation of the Customs final rule, the D.C. Circuit construed the word “confidential” in Exemption 4 and determined that commercial information is “confidential” if disclosure would “cause substantial harm to the competitive position of the person from whom the information was obtained.” National Parks, at 770; see also N.H. Right to Life, 136 S. Ct. at 384. The D.C. Circuit later elaborated that, when applying this test, there was no need to show actual competitive harm and that actual competition and the likelihood of substantial competitive injury sufficed. See Public Citizen Health Research Group v. FDA, 704 F.2d 1280, 1291 (D.C. Cir. 1983); see also N.H. Right to Life, at id. Accordingly, to overcome the strong presumption in favor of disclosure, ruling requesters seeking confidential treatment must prove (1) they actually face competition, and that (2) substantial competitive injury would likely result from disclosure. See Gov't Accountability Project v. FDA, 2016 U.S. Dist. LEXIS 114927, at *15 (D.C. Cir. 2016). Conclusory and generalized allegations of substantial competitive harm are unacceptable and will not support a ruling requester’s efforts to withhold certain information from publication. See Public Citizen Health Research Group, at id.

There is no dispute that Arista faces actual competition and therefore the relevant analysis will focus on the second prong above. See Certain Network Devices, Related Software and Components Thereof, Inv. No. 337-TA-944, Commission Opinion (“Comm’ Op.”) at 58 (June 2016); see also Complainant Cisco Systems, Inc.’s Written Submission in Response to the Commission’s Determination to Review In-Part a Final Initial Determination of a Violation of Section 337 (Brief on Review/Remedy) at 92, Doc ID 580457 (dated May 5, 2016) (“Even assuming that Arista’s products may be used in those spaces [scientific or medical], alternative networking technologies exist, including those supplied by Cisco and others in the industry. See, e.g., Vander Veen Tr. 1220:15–19 (recognizing Arista and Cisco are competitors); CX-10C (Leonard WS) at Q/A 81 (listing other competitors)….Other suppliers can also meet any minor increase in consumer demand caused by the Commission’s remedies. Competitors range from hardware manufacturers to software-defined networking providers. Arista identifies at least seven other companies as its direct competitors: Cisco, Brocade Communications Systems, Dell, Hewlett-Packard, Juniper Networks, IBM, and VMWare.”).

17 

As suggested above, the relevant Customs regulations not only mirror the general presumption in favor of disclosure but also place the burden on the person requesting confidentiality to demonstrate that such information qualifies. See generally FCC v. Schreiber, 381 U.S. 279 (1965) (upholding a rule requiring public disclosure except where the proponents of a request for confidential treatment have established their burden to justify that such information should not be disclosed). Moreover, the provisions of the Customs regulations that place the burden on the ruling requester to establish, during the administrative proceeding, that the information at issue constitutes confidential business information is consistent with the burden the government must satisfy in an action brought against it under FOIA challenging the position an agency has taken not to disclose information pursuant to one or more of the exemptions. See 5 U.S.C. § 552(a)(4)(B); see also United States Department of Justice v. Landano, 508 U.S. 165, 171 (1993) (“The Government bears the burden of establishing that the exemption applies.”).

This line of inquiry is further illuminated by the “general right to inspect and copy public records and documents” that courts have historically recognized, including in patent cases involving highly sensitive information. Apple Inc. v. Samsung Elecs. Co., 727 F.3d 1214, 1221 (Fed. Cir. 2013) (quoting Nixon v. Warner Commc’ns, Inc., 435 U.S. 589 (1978)). The nature of this right, rooted in the common law, is not conditioned on a proprietary interest in the information, see Nixon, at 597, although certainly there are instances where an interest in the information may rise to or approach such a level. Instead, the only interest necessary to compel access to such information is found in the “citizen's desire to keep a watchful eye on the workings of public agencies.” Nixon, at 597-98. This consideration is particularly warranted in regard to the ex parte administrative rulings under 19 C.F.R. § 177 where a person who believes a ruling is incorrect (legally or factually), may seek to have it modified or revoked under 19 C.F.R. § 177.12.

Therefore, a determination whether to grant a request for confidential treatment must balance the likelihood of substantial harm to a competitive position against the need to provide the greatest amount of transparency possible to the agency’s decision- making process. In light of the above, a request for confidential treatment of information submitted in connection with a ruling requested under 19 C.F.R. § 177 faces a strong presumption in favor of disclosure and the person seeking this treatment must overcome that presumption with a request that is narrowly tailored and supported by evidence establishing a likelihood of substantial harm to a competitive position that outweighs the general history and public policy—embedded in FOIA, the common law, and the relevant Customs regulations—favoring transparency and disclosure. To that end, any request for confidential treatment must provide a particularized showing of the specific harm that will result if certain information is disclosed. Broad, unsubstantiated allegations of harm without specific examples articulating reasons for the harm will not suffice.

Notwithstanding the above, it is well recognized by the Federal Circuit and among various federal district courts that, in civil actions arising under the patent laws,

18 

“source code requires additional protections to prevent improper disclosure because it is often a company’s most sensitive and most valuable property . . . (noting that source code might represent a company's ‘most sensitive and confidential property’ and that, in ‘U.S. litigation, extreme measures are ordered to protect [its] confidentiality’). As a result, district courts regularly provide for additional restrictions on discovery to account for the unique characteristics of source code.” Drone Techs., Inc. v. Parrot S.A., 838 F.3d 1283, 1300 n.13 (Fed. Cir. 2016) (internal citations omitted).

Recognizing the significance of source code that is understood in district court actions and Commission investigations under section 337, any source code that forms a part of the administrative record for an ex parte ruling request under 19 C.F.R. § 177 will not be subject to disclosure without permission of submitter. Nevertheless, as at least one district court has acknowledged, “[s]oftware [patent] cases present unique challenges for the parties and the courts because, prior to discovery, plaintiffs usually only have access to the manifestation of the defendants' allegedly infringing source code and not the code itself. From this manifestation, plaintiffs must somehow divine whether the defendants' code infringes. Although defendants vigorously and rightly guard their source code, until plaintiffs have access to it, plaintiffs are typically unable to give highly specified infringement contentions.” American Video Graphics, L.P. v. Electronic Arts, Inc. 359 F. Supp. 2d 558, 560 (E.D. Tex. 2005). Therefore, as noted above, requests for confidential treatment in the context of section 337 exclusion order- based rulings under 19 C.F.R. § 177, especially those involving software running on a particular device, must be narrowly tailored, fully supported by evidence showing a likelihood of substantial harm to a competitive position, and still permit a description of the device’s operation.

Based on the framework above, the information for which Arista has established a likelihood of substantial harm to its competitive position if disclosed (or consists of information that was redacted from the public version of the agency record at the Commission) has been bracketed for redaction from the published ruling but otherwise has been described or identified to the greatest extent allowed throughout this ruling (including, whenever possible, with citations referencing the confidential agency record at the Commission from Inv. No. 337-TA-944). Dark blue brackets (e.g., [XXXX]) signify information that was redacted by the Commission under 19 C.F.R. § 201.6, while red brackets (e.g., [XXXX]) signify information that has been redacted from the submissions to CBP.

C. Law of Patent Infringement

Determining patent infringement requires two steps. Advanced Steel Recovery, LLC v. X-Body Equip., Inc., 808 F.3d 1313, 1316 (2015). The first is to construe the limitations of the asserted claims and the second is to compare the properly construed claims to the accused product. Id. To establish literal infringement, every limitation recited in a claim must be found in the accused product whereas, under the doctrine of equivalents, infringement occurs when there is equivalence between the elements of the accused product and the claimed elements of the patented invention. Microsoft

19  Corp. v. GeoTag, Inc., 817 F.3d 1305, 1313 (Fed. Cir. 2016). One way to establish equivalence is by showing, on an element-by-element basis, that the accused product performs substantially the same function in substantially the same way with substantially the same result as each claim limitation of the patented invention, which is often referred to as the function-way-result test. See Intendis GmbH v. Glenmark Pharms., Inc., 822 F.3d 1355, 1361 (Fed. Cir. 2016).

As for the first step above, “the proper construction of a patent's claims is an issue of Federal Circuit law[.]” Comcast IP Holdings I LLC v. Sprint Communs. Co., 2017 U.S. App. LEXIS 3981, at *11 (Fed. Cir. 2017). Moreover, the ultimate construction of a claim limitation is a legal conclusion, as are interpretations of the patent’s intrinsic evidence (the patent claims, specifications, and prosecution history). UltimatePointer, L.L.C. v. Nintendo Co., 816 F.3d 816, 822 (Fed. Cir. 2016) (citing Teva Pharms. USA, Inc. v. Sandoz, Inc., 135 S. Ct. 831, 841, 190 L. Ed. 2d 719 (2015); see also Multilayer Stretch Cling Film Holdings, Inc. v. Berry Plastics Corp., 831 F.3d 1350, 1357 (Fed. Cir. 2016) (“Claim construction is a question of law, reviewed de novo, but any extrinsic fact-finding by the district court in the course of claim construction is reviewed for clear error.”). The second step to establish infringement involves the comparison of the claims to the accused product, which is a question of fact. Apple Inc. v. Samsung Elecs. Co., Ltd., 839 F.3d 1034, 1040 (Fed. Cir. 2016) (en banc).

The scope and meaning of the asserted claims is determined by looking to the words of the claims themselves, the specification, the prosecution history, and if necessary, any relevant extrinsic evidence. Advanced Steel Recovery, 808 F.3d at 1315-17. Moreover, claim terms are generally given their ordinary and customary meaning, which is the meaning they would have to a person of ordinary skill in the art at the time of the invention. Poly-America, L.P. v. API Indus., Inc., 839 F.3d 1131, 1136 (Fed. Cir. 2016) (citing Phillips v. AWH Corp., 415 F.3d 1303, 1312-13 (Fed. Cir. 2005) (en banc)). “Importantly, the person of ordinary skill in the art is deemed to read the claim term not only in the context of the particular claim in which the disputed term appears, but in the context of the entire patent, including the specification.” UltimatePointer, 816 F.3d at 822 (quoting Phillips, 415 F.3d at 1313); see also Retractable Techs., Inc. v. Becton, 653 F.3d 1296, 1305 (Fed. Cir. 2011) (“It is axiomatic that the claim construction process entails more than viewing the claim language in isolation.”). In particular, “the specification ‘is always highly relevant to the claim construction analysis. Usually, it is dispositive; it is the single best guide to the meaning of a disputed term.’” Phillips, at 1313 (quoting Vitronics Corp. v. Conceptronic, Inc., 90 F.3d 1576, 1582 (Fed. Cir. 1996)). However, while claims are read in view of the specification for which they are a part, it is improper to read limitations from the embodiments in the specification into the claims. Hill-Rom Servs. v. Stryker Corp., 755 F.3d 1367, 1371 (Fed. Cir. 2014).

Accordingly, the ordinary and customary meaning of the claim language generally controls unless the patentee acts as his own lexicographer and provides a special definition for a particular claim term or the patentee disavows the ordinary scope of a claim term either in the specification or during prosecution. InterDigital Communs.,

20 

LLC v. U.S. Int’l Trade Comm’n, 690 F.3d 1318, 1324 (Fed. Cir. 2012); see also Hill- Rom Servs., 755 F.3d at 1371 (“There are only two exceptions to this general rule: 1) when a patentee sets out a definition and acts as his own lexicographer, or 2) when the patentee disavows the full scope of the claim term either in the specification or during prosecution.") (quoting Thorner v. Sony Computer Entm't Am. LLC, 669 F.3d 1362, 1365 (Fed. Cir. 2012)).

The standards for finding lexicography and disavowal are "exacting." Pacing Techs., LLC v. Garmin Int'l, Inc., 778 F.3d 1021, 1024 (Fed. Cir. 2015). “To act as a lexicographer, a patentee must ‘clearly set forth a definition of the disputed claim term’ and ‘clearly express an intent to define the term.’” Id. (quoting Thorner, 669 F.3d at 1365. “Similarly, disavowal requires that ‘the specification [or prosecution history] make[] clear that the invention does not include a particular feature.’” Id. (quoting SciMed Life Sys. Inc. v. Advanced Cardiovascular Sys., Inc., 242 F.3d 1337, 1341 (Fed. Cir. 2001). For example, disavowal or disclaimer has been found based on “clear and unmistakable statements by the patentee that limit the claims, such as ‘the present invention includes . . .’ or ‘the present invention is . . .’ or ‘all embodiments of the present invention are . . . .’” Id. (citing Regents of Univ. of Minn. v. AGA Med. Corp., 717 F.3d 929, 936 (Fed. Cir. 2013); Honeywell Int'l, Inc. v. ITT Indus., Inc., 452 F.3d 1312, 1316-19 (Fed. Cir. 2006); SciMed Life Sys., Inc., 242 F.3d at 1343-44).

Ultimately, "[t]he only meaning that matters in claim construction is the meaning in the context of the patent," Trs. of Columbia Univ. v. Symantec Corp., 811 F.3d 1359, 1363 (Fed. Cir. 2016), and “[l]egal error arises when a court relies on extrinsic evidence that contradicts the intrinsic record.” Ruckus Wireless, Inc. v. Innovative Wireless Solutions, LLC, 824 F.3d 999, 1003 (Fed. Cir. 2016) (citing Lighting Ballast Control LLC v. Philips Elecs. N. Am. Corp., 790 F.3d 1329, 1338 (Fed. Cir. 2015), cert. denied by Universal Lighting Techs., Inc. v. Lighting Ballast Control, LLC, 136 S. Ct. 1226; 194 L. Ed. 2d 185; 2016 U.S. LEXIS 1155 (2016). Therefore, "[t]he construction that stays true to the claim language and most naturally aligns with the patent's description of the invention will be, in the end, the correct construction." Papst Licensing GmbH & Co. KG v. Fujifilm Corp., 778 F.3d 1255, 1261 (Fed. Cir. 2015) (quoting Renishaw PLC v. Marposs Societa' per Azioni, 158 F.3d 1243, 1250 (Fed. Cir. 1998), adopted by Phillips, 415 F.3d at 1316).

Lastly, only those claim terms that are disputed must be construed and only to the extent necessary to resolve the dispute. See O2 Micro Int'l Ltd. v. Beyond Innovation Tech. Co., 521 F.3d 1351, 1362 (2008) (“We, however, recognize that district courts are not (and should not be) required to construe every limitation present in a patent's asserted claims”). Rather, “[c]laim construction is a matter of resolution of disputed meanings and technical scope, to clarify and when necessary to explain what the patentee covered by the claims, for use in the determination of infringement.” Id. (quoting U.S. Surgical Corp. v. Ethicon, Inc., 103 F.3d 1554, 1568 (Fed. Cir. 1997); see also Vanderlande Indus. Nederland BV v. U.S. Int’l Trade Comm’n, 366 F.3d 1311, 1323 (Fed. Cir. 2004).

21 

D. Section 337 Exclusion Order Administration

Under section 337 of the Tariff Act of 1930, as amended, 19 U.S.C. § 1337, the Commission has authority to conduct investigations into imported articles that allegedly infringe United States patents and impose remedies if the accused articles are found to be infringing. See 19 U.S.C. § 1337(a)(1)(B), (b)(1), (d), (f), (i). 19 U.S.C. § 1337(d), in particular, empowers the Commission to direct the exclusion from entry of articles found to infringe the asserted patents. When the Commission determines there is infringement and thus a violation of section 337, it generally issues one of two types of exclusion orders: (1) a limited exclusion order or (2) a general exclusion order. See Fuji Photo Film Co., Ltd. v. U.S. Int’l Trade Comm’n, 474 F.3d 1281, 1286 (Fed. Cir. 2007).

Both types of orders direct CBP to bar infringing products from entering the country. See Yingbin-Nature (Guangdong) Wood Indus. Co. v. U.S. Int’l Trade Comm’n, 535 F.3d 1322, 1330 (Fed Cir. 2008). “A limited exclusion order is ‘limited’ in that it only applies to the specific parties before the Commission in the investigation. In contrast, a general exclusion order bars the importation of infringing products by everyone, regardless of whether they were respondents in the Commission's investigation.” Id. A general exclusion order is appropriate only if two exceptional circumstances apply. See Kyocera Wireless Corp. v. U.S. Int’l Trade Comm’n, 545 F.3d 1340, 1356. A general exclusion order may only be issued if (1) “necessary to prevent circumvention of a limited exclusion order,” or (2) “there is a pattern of violation of this section and it is difficult to identify the source of infringing products.” 19 U.S.C. § 1337(d)(2); see also Kyocera, 545 F.3d at 1356 (“If a complainant wishes to obtain an exclusion order operative against articles of non-respondents, it must seek a GEO [general exclusion order] by satisfying the heightened burdens of §§ 1337(d)(2)(A) and (B).”).

In addition to the action taken above, the Commission may issue an order under 19 U.S.C. § 1337(i) directing CBP to seize and forfeit articles attempting entry in violation of an exclusion order if their owner, importer, or consignee previously had articles denied entry on the basis of that exclusion order and received notice that seizure and forfeiture would result from any future attempt to enter articles subject to the same. An exclusion order under § 1337(d)—either limited or general—and a seizure and forfeiture order under § 1337(i) apply at the border only and are operative against articles presented for customs examination or articles conditionally released from customs custody but still subject to a timely demand for redelivery. See 19 U.S.C. §§ 1337(d)(1) (“The Commission shall notify the Secretary of the Treasury of its action under this subsection directing such exclusion from entry, and upon receipt of such notice, the Secretary shall, through the proper officers, refuse such entry.”) and (i)(3) (“Upon the attempted entry of articles subject to an order issued under this subsection, the Secretary of the Treasury shall immediately notify all ports of entry of the attempted importation and shall identify the persons notified under paragraph (1)(C).”) (emphasis added); see also ClearCorrect Operating, LLC v. U.S. Int’l Trade Comm’n, 810 F.3d 1283, 1295 (Fed. Cir. 2015); rehearing en banc denied by Clearcorrect Operating v. ITC, 819 F.3d 1334 (Fed. Cir. 2016) (“This section [1337(i)] permits the Commission to

22 

exclude ‘articles’ from importation into the United States; however, it is difficult to see how one could physically stop electronic transmissions at the borders under the current statutory scheme…. A construction of the term ‘articles’ that includes electronically transmitted digital data is also not reasonable when applied to Section 337(i)(3). This section reads, ‘[u]pon the attempted entry of articles subject to an order issued under this subsection, the Secretary of the Treasury shall immediately notify all ports of entry of the attempted importation and shall identify the persons notified under paragraph (1)(C).’ Not only can an electronic transmission not be subject to an ‘attempted entry’ through a ‘port of entry,’ it also cannot be intercepted at a ‘port of entry’ as contemplated in the statute.”) (emphasis added).

Significantly, unlike district court injunctions, the Commission can issue a general exclusion order that broadly prohibits entry of articles that infringe the relevant claims of an asserted patent without regard to whether the persons importing such articles were parties to, or were related to parties to, the investigation that led to issuance of the general exclusion order. See Vastfame Camera, Ltd. v. U.S. Int’l Trade Comm’n, 386 F.3d 1108, 1114 (Fed. Cir. 2004). The Commission also has recognized that even limited exclusion orders have broader applicability beyond just the parties found to infringe during an investigation. See Certain GPS Devices and Products Containing Same, Inv. No. 337-TA-602, Comm’ Op. at 17, n. 6, Doc ID 317981 (Jan. 2009) (“We do not view the Court’s opinion in Kyocera as affecting the issuance of LEOs [limited exclusion orders] that exclude infringing products made by respondents found to be violating Section 337, but imported by another entity. The exclusionary language in this regard that is traditionally included in LEOs is consistent with 19 U.S.C. § 1337(a)(1)(B) - (D) and 19 U.S.C. § 1337(d)(1).”).

However, this is not the only difference between district court injunctions under section 283 of the Patent Act and Commission exclusion orders under section 337 of the Tariff Act. For example, the traditional test for injunctive relief, used by district courts in accordance with the Supreme Court’s eBay Inc. v. MercExchange, L.L.C., 547 U.S. 388 (2006), “does not apply to Commission remedy determinations under Section 337.” Spansion, Inc. v. U.S. Int’l Trade Comm’n, 629 F.3d 1331, 1359 (Fed. Cir. 2010). This difference between exclusion orders under Section 337 and injunctions under the Patent Act follows “the long-standing principle that importation is treated differently than domestic activity.” Id. at 1360 (citing United States v. 12 200-Ft. Reels of Super 8mm Film, 413 U.S. 123 (1973) (“Import restrictions and searches of persons or packages at the national borders rest on different considerations and different rules of constitutional law from domestic regulations. The Constitution gives Congress broad, comprehensive powers ‘to regulate Commerce with foreign Nations.’ Art. I, § 8, cl. 3.”); see also Buttfield v. Stranahan, 192 U.S. 470 (1904) (“As a result of the complete power of Congress over foreign commerce, it necessarily follows that no individual has a vested right to trade with foreign nations, which is so broad in character as to limit and restrict the power of Congress to determine what articles of merchandise may be imported into this country and the terms upon which a right to import may be exercised. This being true, it results that a statute which restrains the introduction of particular goods into the

23 

United States from considerations of public policy does not violate the due process clause of the Constitution.”) (emphasis added).

Moreover, in the district courts, the criteria for adjudicating a violation of an injunction against future infringement by a party whose legacy products have already been adjudged to infringe requires application of the colorable differences test. TiVo Inc. v. Echostar Corp., 646 F.3d 869, 876 (Fed. Cir. 2011) (en banc). Under this test, if new or redesigned products are so different from the product previously found to infringe (focusing not on differences randomly chosen between the two but on the specific features of the product found to infringe in the earlier infringement trial), such that they raise a fair ground of doubt as to the wrongfulness of the defendant's conduct, the new or redesigned products are deemed to be more than colorably different from the legacy one adjudged to be infringing and violation of the injunction at this point would not be the appropriate consideration. Instead, a new trial would be required to litigate the infringement question. Id. at 881-83. The initial inquiry under the colorable differences test, therefore, focuses on whether the modification in question is significant. Id.

Exclusion orders, however, apply to any articles—new, modified, or otherwise— that are “covered by” the patent claims included in the exclusion order. See Certain Optical Disk Controller Chips and Chipsets, Inv. No. 337-TA-506, Comm’ Op. at 56-57, USITC Pub. 3935, Doc ID 287263 (July 2007) ("The Commission's long-standing practice is to direct its remedial orders to all products covered by the patent claims as to which a violation has been found, rather than limiting its orders only to those specific models selected for the infringement analysis...[W]hile individual models may be evaluated to determine importation and infringement, the Commission's jurisdiction extends to all models of infringing products that are imported at the time of the Commission's determination and to all such products that will be imported during the life of the remedial orders."). The Commission has confirmed that this requires CBP to employ the traditional two-step test for patent infringement, and not the colorable differences test from TiVo, when administering a Commission exclusion order under section 337 to determine whether an imported article, or one at issue in a ruling request, is within the scope of that exclusion order. See Certain Sleep-Disordered Breathing Treatment Systems and Components Thereof, Inv. No. 337-TA-879, Advisory Opinion at 10, n. 2, Doc ID 539875 (Aug. 2014); see also Certain Erasable Programmable Read Only Memories, Inv. No. 337-TA-276 Enforcement Proceeding Comm’ Op. at 11, Doc ID 43536 (Aug. 1991) (“The Commission has always issued its orders in terms of ‘infringing’ products, and has always intended them, as in this case, to prohibit to future importation or sale of products which were not specifically adjudged infringing in the violation proceeding, but do in fact infringe. The Commission has consistently issued exclusion orders coextensive with the violation of section 337 found to exist. Thus, in cases where the violation found involves infringement of patent claims, the Commission has consistently ordered the exclusion of articles which infringe the relevant patent claims.”) (emphasis added).

24 

Notwithstanding the above, and as noted in the revoked ruling, there is a question whether the term “covered by” in an exclusion order requires CBP, during its administration, to determine whether there is infringement of the relevant patent claims under the doctrine of equivalents. As mentioned above, “a product or process that does not literally infringe the express terms of a patent claim may nonetheless be found to infringe if there is 'equivalence' between the elements of the accused product or process and the claimed elements of the patented invention." Duramed Pharms., Inc. v. Paddock Labs., Inc., 644 F.3d 1376, 1390 (Fed. Cir. 2011) (quoting Warner-Jenkinson Co. v. Hilton Davis Chem. Co., 520 U.S. 17 (1997).

However, this doctrine has never been applied by CBP during its administration of an exclusion order to refuse entry of an article (other than, for example, where the Commission would a violation of section 337, during the underlying investigation, by applying the doctrine) or to find, through a determination under 19 C.F.R. §§ 174 or 177, that an article is subject to an exclusion order due to application of the doctrine.

Moreover, despite the revocation of CBP HQ Ruling H277148, the concerns animating its analysis remain and call into question whether it would be appropriate for CBP, in the first instance, to apply the doctrine when administering an exclusion order. Those concerns, as referenced in the revoked ruling, include the limitations with the ex parte nature of CBP’s decision-making under the current regulatory framework that is less than ideally suited for application of the doctrine; the unique features of section 337 highlighted above and below (particularly in respect of the shifted burden to establish admissibility following issuance of an exclusion order); the lack of a Commission record with findings of fact on the precise issue, especially with regard to the function-way- result test; and the resort, that inevitably would follow, to prosecution history estoppel as a defense to infringement or legal limitation on the application of the doctrine. Furthermore, it is not clear whether the Commission intends for CBP to apply this doctrine when administering of an exclusion order.

Due to the last concern noted above, CBP submitted the question to the Commission whether CBP should apply to the doctrine when administering an exclusion order. Specifically, CBP asked whether an exclusion order issued by the Commission requires CBP to apply the doctrine, regardless of whether the Commission found a violation of section 337 under the doctrine during the investigation that gave rise to the exclusion order; or whether the Commission does not authorize CBP to apply the doctrine in any instance; or whether the question of agency action involving this doctrine is committed to CBP’s discretion in the sound administration of the Customs and related laws, including section 337; or, lastly, whether the Commission envisions another position that is not illuminated above. CBP understands that this question is still under consideration at the Commission.

Accordingly, for the reasons above and except as otherwise directed, CBP will consider the question regarding application of the doctrine as being committed to its discretion and will apply the doctrine in an appropriate manner to carry out the sound administration of the Customs and related laws. See Heckler v. Chaney, 470 U.S. 821,

25 

830 (1985) (explaining what it means for an action to be "committed to agency discretion by law" and employing a test where certain agency decisions declining to initiate enforcement actions were held to be presumptively unreviewable as a category); see also Webster v. Doe, 486 U.S. 592, 608 (1988) (“The ‘no law to apply’ test can account for the nonreviewability of certain issues, but falls far short of explaining the full scope of the areas from which the courts are excluded. For the fact is that there is no governmental decision that is not subject to a fair number of legal constraints precise enough to be susceptible of judicial application -- beginning with the fundamental constraint that the decision must be taken in order to further a public purpose rather than a purely private interest; yet there are many governmental decisions that are not at all subject to judicial review.”).

As such, CBP will not extend the doctrine of equivalents when administering an exclusion order pursuant to section 337 except in two instances. The first is where the Commission found a violation of section 337, during the underlying investigation, through infringement under the doctrine. The second is, in those cases when the Commission has found only literal infringement of the asserted patents, where a respondent identified in an exclusion order fails to show a prima facie case that the doctrine does not apply to its new or modified article under the “function-way-result” or “insubstantial differences” test based on the administrative record before CBP.

This approach is consistent with the Commission’s pronouncement above that it “has consistently issued exclusion orders coextensive with the violation of section 337 found to exist.” Certain Erasable Programmable Read Only Memories, Inv. No. 337- TA-276 Enforcement Proceeding Comm’ Op. at 11, Doc ID 43536 (Aug. 1991) (emphasis added). Where the violation found to exist includes an infringement finding under the doctrine, CBP will apply that doctrine during its administration of the exclusion order. However, in those investigations where the Commission only finds literal infringement, CBP will extend the doctrine under the test established above. This approach is also consistent with other Commission precedent regarding the scope and reach of its exclusion orders. See Certain Ground Fault Circuit Interrupters and Products Containing the Same, Inv. No. 337-TA-615, Comm’ Op. at 27-28, Doc ID 321819 (March 2009) (“The Commission's practice is consistent with Federal Circuit law, which provides that the Commission's infringement determinations with respect to the adjudicated products are effective for the purposes of the exclusion order against different models presented for importation at a future date if there is a ‘close identity between the relevant features of an accused device and the device determined to be infringing.’ Correspondingly, the exclusion order would not apply to products not adjudicated to be infringing, and not having such a ‘close identity,’ thus alleviating respondents' concerns, unless infringement is established by other means.”) (emphasis added) (internal citations omitted).

CBP, therefore, will not extend the doctrine of equivalents when administering an exclusion order except in the two instances described above, subject to modification or revocation of this position pursuant to 19 C.F.R. § 177.12(d)(1) and in furtherance of an order, instruction or other policy decision from another governmental agency.

26  Lastly, despite the well-established principle that “the burden of proving infringement generally rests upon the patentee,” Medtronic, Inc. v. Mirowski Family Ventures, LLC, 134 S. Ct. 843; 187 L. Ed. 2d 703; 2014 U.S. LEXIS 788 (2014), the Commission has held that Medtronic is not controlling precedent and does not overturn its longstanding practice of placing the burden of proof on the party who, in light of the issued exclusion order, is seeking to have an article entered for consumption. See Certain Sleep- Disordered Breathing Treatment Systems and Components Thereof, Inv. No. 337-TA- 879, Advisory Opinion at 6-11. In particular, the Commission has noted that “[t]he Federal Circuit has upheld a Commission remedy which effectively shifted the burden of proof on infringement issues to require a company seeking to import goods to prove that its product does not infringe, despite the fact that, in general, the burden of proof is on the patent to prove, by a preponderance of the evidence, that a given article does infringe the patent in question.” Certain Integrated Circuit Telecommunication Chips, Inv. No. 337-TA-337, Comm’ Op. at 21, n. 14, USITC Pub. 2670, Doc ID 217024 (Aug. 1993), citing Sealed Air Corp. v. U.S. Int’l Trade Comm’n, 645 F.2d 976, 107-08 (C.C.P.A. 1981) (emphasis in original).

This approach is supported by Federal Circuit precedent. See Hyundai Elecs. Indus. Co. v. U.S. Int'l Trade Comm'n, 899 F.2d 1204, 1210 (Fed. Cir. 1990) (“Indeed, we have recognized, and Hyundai does not dispute, that in an appropriate case the Commission can impose a general exclusion order that binds parties and non-parties alike and effectively shifts to would-be importers of potentially infringing articles, as a condition of entry, the burden of establishing noninfringement. The rationale underlying the issuance of general exclusion orders—placing the risk of unfairness associated with a prophylactic order upon potential importers rather than American manufacturers that, vis-a-vis at least some foreign manufacturers and importers, have demonstrated their entitlement to protection from unfair trade practices—applies here [in regard to a limited exclusion order] with increased force.”) (emphasis added, internal citation omitted).

Accordingly, the burden is on Arista to establish that the network devices at issue do not infringe one or more of claims 1, 2, 8-11, and 17-19 of the '537 patent; claims 6, 7, 20, and 21 of the '592 patent; or claims 5, 7, 45, and 46 of the '145 patent.

E. Admissibility of Redesigned Products in Light of the `537 Patent

1. Construction of Claim Limitations at Issue

Neither party has argued to CBP or the Commission that this dispute turns on the claim construction of any limitation from the `537 patent. See Complainant Cisco Systems, Inc.'s Prehearing Brief and Statement at 15 (“Neither party has proposed claim construction in this Enforcement Proceeding.”). As such, CBP will not construe any of the claim limitations at issue and will instead give them their ordinary and customary meaning. See O2 Micro Int'l Ltd. v. Beyond Innovation Tech. Co., 521 F.3d 1351, 1362 (2008) (“We, however, recognize that district courts are not (and should not be) required to construe every limitation present in a patent's asserted claims” and that

27 

“[c]laim construction is a matter of resolution of disputed meanings and technical scope, to clarify and when necessary to explain what the patentee covered by the claims, for use in the determination of infringement.”) (quoting U.S. Surgical Corp. v. Ethicon, Inc., 103 F.3d 1554, 1568 (Fed. Cir. 1997); see also Vanderlande Indus. Nederland BV v. U.S. Int’l Trade Comm’n, 366 F.3d 1311, 1323 (Fed. Cir. 2004).

2. Removal of [ ] From Legacy Products Found to Infringe

A starting point for addressing whether Arista’s network devices with the EOS redesign are covered by the `537 patent, and therefore are subject to exclusion from entry under the LEO, is with the question whether, as Arista states, it has removed the “[ ]” that were found during the investigation at the Commission to meet the claim limitation requiring a “management registration request.” See Arista's Prehearing Brief at 8 (“But in the redesigned EOS, agents never transmit a [ ], so the ‘management registration request’ limitation cannot be met. Ex. 1 (McKusick Decl) ¶ 15.”); see also id. at 26 (“To eliminate any doubt that its software does not infringe the ’537 patent, Arista removed [ ] sent from agents to Sysdb entirely from EOS. RX-5133C (Duda RWS) at Q/A 17-18; RX-5131C (Sweeney RWS) at Q/A 52-56.”).

Cisco, while pointing out that Arista raises one primary non-infringement argument (i.e., that its redesign no longer has a managing subsystem that transmits a management registration request), does not concede that Arista has removed [ ] from the redesign EOS. See Cisco Pre-Proceeding Submission (dated February 2, 2017) at 2 (“But in reality, Arista did not remove the [ ]; it simply attempted to disguise it, by maintaining that functionality under another name— ‘ProcMgr.’ Although Arista made a superficial change to the [ ] procedure, the message sent from ProcMgr is indisputably still a [ ]….”); see also Cisco Systems, Inc.'s Prehearing Brief and Statement at 55 (“[I]n reality, Arista did not remove the [ ]; it simply attempted to disguise it, by maintaining that functionality as an agent-specific process under another name-[ ] See, e.g., CX-5002C (Almeroth WS) at Q/A 115-118, 127-131, 213, 220. Although Arista made a superficial change to the [ ] procedure, the message sent by the agent-specific [ ] process is indisputably still a [ ] request, as explained below in Sections VI.B.1 and VI.B.2.b.”).

Moreover, Cisco’s belief that Arista has not removed [ ] from the redesign EOS is reflected by the continued use in Cisco’s submissions, both to CBP and the Commission, of the term “[ ]” when referring to the communication from ProcMgr to Sysdb that Arista refers to as the [ ] command. See Cisco Systems, Inc.'s Prehearing Brief and Statement (“ProcMgr calls the [ ] for that specific agent, which sends a [ ] message to Sysdb – which Arista now calls the ‘[ ] command’ – containing the [ ] for which a [ ] is being requested.”) (emphasis added); see also Cisco Systems, Inc.'s Prehearing Brief and Statement at 66 (“Likewise, there is

28 

no question that the agent-specific processes transmit ‘management registration requests’ in the form of [ ] for each particular agent in Arista's system.”); see also Cisco Pre-Proceeding Submission at 13 (“There is also no dispute that ProcMgr issues [ ] for each individual agent, by sending a message (the “[ ] command”) with the [ ] to Sysdb, which Sysdb uses to put [ ] in place); see also Cisco Pre-Proceeding Submission at 5 (“Likewise, as Cisco explained in its opening submission, the agent-specific processes in ProcMgr transmit management registration requests in the form of [ ] for each particular agent in Arista’s system.”).

Moreover, according to Cisco, Arista attempts to hide the fact that “[ ]” are still present in the redesigned EOS through use of the term “[ ] command.” See Cisco Systems, Inc.'s Prehearing Brief and Statement at 67, FN 8 (“Arista's attempts to obscure its continued infringement is evidenced by the fact that Arista fabricated the term ‘[ ]’ in an attempt to portray its [ ] as a [ ], and to conceal the undisputed facts that the request is sent by an agent- specific process to Sysdb to request [ ] to be put in place for that particular agent. In reality, there is no ‘[ ] command’ in the EOS source code. See, e.g., CX-5656C (Arista's Response to Cisco's Third Set of Requests For Admission) at RFA No. 287; CX-5013C (Sweeney Dep. Designations) at 229:1-5. The name ‘[ ]’ also does not appear to have ever been used in internal engineering discussions at Arista, either, based on Arista's documentations. See generally, e.g., CX-5043C (‘[ ]’); CX-5042C (‘[ ]’)).”.

Arista responds that “Cisco resists using the name ‘[ ],’ but does not dispute that the [ ] is run to send a message to [ ] that includes the [ ].” See Arista's Prehearing Brief at 52, FN 12. This is consistent with the testimony of Arista’s expert witness, Dr. McKusick.

[

]

McKusick De. Tr. at 74:3-9.

At one point, Cisco even suggests that Arista merely transferred the code from one place in the legacy EOS to another place in the redesign EOS. Cisco Systems, Inc.'s Prehearing Brief and Statement at 103 (“As explained above, when Arista redesigned its EOS software, it moved a portion of the code responsible for sending [ ] to a process called ‘[ ].’ See, e.g., Sections VI.A and VI.B.”). However, those prior sections cited in Cisco’s brief only make reference to the

29  transfer of functionality instead of the transfer of code from one place in the operating system to another. This view regarding the transfer of functionality but not transfer of actual code is consistent with the testimony of Cisco’s expert witness, Dr. Almeroth.

Q In theredesigned EOS generally, so just asking general questions again, do you agree with me that you don't identify in your report any source code that was transferred from -- from what used to be in the agent in the legacy EOS, to ProcManager in the redesigned EOS? No actual source code was transferred? A I don’t recall that I’ve offered that opinion in the report. I think the report speaks for itself. I think what I describe is, is moving the functionality. But again, I think the report speaks for itself.

Almeroth De. Tr. at 74:3-9.

Despite Cisco’s one reference to the transfer of code and the broader refrain that [ ] were not removed from the EOS, the evidence supports a finding that Arista has removed the [ ] found at the Commission to practice the “management registration request” limitation.

As stated in Arista’s prehearing brief to the ALJ, in the legacy EOS, agents that needed to [ ] to Sysdb would make a request to Sysdb to [ ]. See RX-5129C (McKusick RWS) at Q/A 18-20, 82-86, 139-143, 216-218, 261. Each agent had a [ ] associated with its connection to Sysdb, and Sysdb had a separate instance of the same type of [ ] associated with the connection to each individual agent. Id. The

]. Id. Each entry in the [

]. After an agent in the legacy EOS created an entry [

]. Id. When the [

]. Id. Cisco’s prehearing brief does not contradict the description above. See Cisco Systems, Inc.'s Prehearing Brief and Statement at 24-54.

Arista’s expert witness has confirmed that the [ ] (specifically, the [

30 

]) are no longer present in the redesign EOS. See Arista Pre- Proceeding Submission, Ex. 01 – McKusick Declaration, ¶ 15. Cisco has not provided evidence that calls into question Arista’s claim that agents no longer have the ability to send [ ].

Accordingly, it is not disputed that, in the legacy EOS, it was the transmission by the EOS agent to Sysdb of the [

] Cisco accused and the Commission found to satisfy the claim limitation requiring a “management registration request” and that, in the redesign EOS, these [ ] have been removed from the redesign EOS.

For the reasons above, the administrative record supports a finding that the [ ] found to infringe at the Commission have been removed from the EOS and, as such, no longer are sent from the agents to Sysdb. Furthermore, as described below, the record supports Arista’s characterization that, instead of the [

]; Sysdb, upon receiving the [ ] command, [ ]. See Arista's Prehearing Brief at 26; see also RX-5129C (McKusick RWS) at Q/A 27-41, 53-66, 147-205; RX-5131C (Sweeney RWS) at Q/A 43-142.

3. Issuance of Management Registration Request and Actual External Management

As found above, Arista removed the [ ] found to infringe at the Commission and, moreover, has replaced the legacy EOS with a redesign where the process that sends the [ ] command is not the process that externally manage router configuration data. Arista points out, and Cisco does not contest, that the claimed “managing subsystem” in the `537 patent is responsible for performing two of the requirements found in the claims. First, it is responsible for sending a management registration request to the database subsystem and, second, it is responsible for externally managing router configuration store in the database subsystem. See Arista's Prehearing Brief at 15 (“And, most importantly, the requirement that the management request come from the managing subsystem is a central requirement of every claim Arista was found to infringe. For example, claim 19…explicitly requires that the managing subsystem send the management request to the database in both of its main limitations”) (as underlined in red below).

31 




JX-0001 (’537 Patent) at claim 19. Arista points out that the same is true for claims 1 and 10:





Id.

As Arista puts it, “[t]he ‘managing subsystem’ of the ’537 patent has two fundamental and inescapable characteristics recited in the claims: 1. It performs external management; and 2. It ‘transmits’ or ‘issues’ the management registration request[.]”) Arista's Prehearing Brief at 45; see also id. at 47 (“Taken together, the claim language is unmistakable: to infringe, there must be a managing subsystem— meaning a subsystem that is or requests to be an external manager of router

32 

configuration data in the database subsystem—that issues or transmits the management registration request.”).

In light of this requirement in the claims, Arista’s view is that infringement cannot lie given that there is no dispute between the parties that ProcMgr does not externally manage router configuration data and that agents do not send the “[ ]” command. See Arista's Prehearing Brief at 50 (“This testimony [that neither ProcMgr nor [ ] is capable of [ ] is case-dispositive. The only things sending alleged management registration requests in the redesign are ProcMgr and [ ], and both parties agree that neither is a managing subsystem. The express language of the claims is not met, and this is not a close call.”); see also id. at 53 (“Here too, Cisco’s concessions [that the [ ] is run by ProcMgr, is not run by an agent, and is run prior to the [ ] are case-dispositive. It admits that the only things that could externally manage data in Sysdb are agents, but agents never send [ ] commands, and never send [ ] or external management requests of any kind.”).

Based on the above, Arista has established, and Cisco does not controvert, that in the redesign EOS ProcMgr sends the [ ] command to Sysdb without any involvement or assistance of the agent that externally manages data. Accordingly, in the redesign, there is no process that both sends a communication that would satisfy the management registration request limitation and also externally manages router configuration data. As such, Arista has established admissibility of its redesign EOS since it is not covered by the claims of the `537 patent unless there are reasons to consider ProcMgr and the agents, together, to comprise the claimed “managing subsystem.”

Cisco argues that the separate processes, ProcMgr and an agent (or, as Cisco specifically focuses its analysis, the agent-specific [ ] in ProcMgr and the agent) should be considered together and that, when considered together, they satisfy the managing subsystem limitation because—through the two separate processes— there is the transmission of a communication that qualifies as a management registration request and also the external management of router configuration data. The arguments raised by each side concerning this issue are analyzed below to see if there are sufficient reasons that the agent-specific [ ] and the agent should be considered in combination or otherwise viewed as working together.

4. Reasons Whether to Consider ProcMgr and an Agent Together

As an initial matter, it is important to note that the `537 patent and specification never expressly claims or depicts a router or client subsystem that is comprised of multiple processes. Instead, the patent claims and specification refers to or shows a single subsystem that is configured as a managing subsystem and coupled for communication to the claimed centralized database subsystem (“sysDB”). In particular, as Arista notes, the `537 patent, when describing the prior art internetwork operating

33 

systems for router devices, refers to “subsystem components which provide specific functions or routines provided by the routing device.” JX-0001 (’537 patent) at 1:14-18. As for the `537 patent, the claimed router or client subsystems are illustrated in Figure 2, as depicted below.




In this depiction, Figure 2 also shows the managing subsystem as a single subsystem, from which among these plurality of router or client subsystem is configured a managing subsystem. As Arista points out, the patent never describes two or more subsystems as being combined to form a managing subsystem. JX-0001 (’537 patent) at claims 1, 10, 19.

Instead, Figure 4 of the `537 patent shows that a router or client subsystem configured as a managing subsystem also consists of a single process that includes certain component parts.




JX-0001 (’537 Patent) at Fig. 4.

34 

Moreover, Cisco states that “the agent-specific [ ] in ProcMgr is called . . . on an agent-by-agent basis for the exclusive purpose of sending a [ ] and [ ] for the associated agent.” See Cisco Post-Proceeding Submission (dated February 23, 2017) at 5. Cisco further states that “it is undisputed that the agent-specific processes in ProcMgr for establishing [ ] and [ ] are performed individually for each agent in the system.” Id. at 6 (emphasis in the original).

While ProcMgr calls certain functions, i.e., the [ ] and the [ ], on an agent-by-agent basis, or for each agent that it is starting, ProcMgr is nonetheless one continuous process. See Arista Pre-Proceeding Submission (dated February 2, 2017) at 26 (“The purpose of process IDs . . . is to identify separate processes” and “. . . the process ID of the ProcMgr (parent process) remains the same. It does not change with respect to each child process it starts. ProcMgr is designed to be a process that is always running without interruption.”) (citations omitted)(emphasis added); see also Arista Pre-Proceeding Submission (dated February 2, 2017) at 13-14 (in which the Google Chrome example shows multiple instances of chrome.exe “running at a given moment, each of which is a separate process.”)(emphasis added)). Even though there are multiple agents or child processes, there is only one ProcMgr or parent process for these agents. Id. The same functions are being called by the same process, albeit repeatedly with a different variable plugged in each time (i.e., the [ ]), for [ ]. There is no basis for breaking apart the one ProcMgr process into multiple pieces or functions and then appending these functions to a separate process, i.e., the agent or child process, to satisfy the “managing subsystem” limitation. See e.g., Arista Pre- Proceeding Submission (dated February 2, 2017) at 46 (“It is inappropriate to consider the union of just one aspect of a subsystem with another subsystem. . . .”) (emphasis added); see also id. at 44 (“The proposed combination is akin to combining Google Chrome’s web browser functionality with Microsoft Word’s spell checker functionality, and labeling that collection a single subsystem, even though Chrome and Word exist as two separate programs and run as separate processes.”).

Cisco also argues that “[n]othing in the ‘537 patent excludes the use of multiple processes or software routines to implement a managing subsystem,” but this does not address why it would be appropriate to break out certain functions (i.e., the [ ] and the [ ]), which are part of a process (i.e., ProcMgr) and can only be used by that process, from that process to combine with a separate process (i.e., an agent) to form a “managing subsystem.” See Cisco Pre- Proceeding Submission (dated February 2, 2017) at 23. As Arista implies the “managing subsystem” cannot be satisfied by “an arbitrary collection of functionalities from disparate programs that run as separate processes.” Arista Post-Proceeding Submission (dated February 23, 2017) at 24; see also Arista Letter (dated September 19, 2016) at 7 (subsystems are not “an amorphous collection of functionally disparate components that collectively lead to some result.”). Moreover, Arista’s legacy products also had a ProcMgr, which started and stopped agents (although it did not issue any

35 

[ ]) on an agent-by-agent basis, but the ProcMgr in the legacy products was not claimed to be part of the "managing subsystem." See Arista Post-Proceeding Submission (dated February 23, 2017) at 16 (". . . ProcMgr started and stopped specific agents in the legacy EOS, and when it did so, it called the [ ] for the specific agent it was starting, just as it does in the redesigned EOS.").

Cisco also attempts to liken the ProcMgr in Arista's redesigned product to the EntityManager in Arista's legacy products, but there is a notable distinction between the two, namely EntityManager is a library and ProcMgr is a process. See Cisco Post­ Proceeding Submission (dated February 23, 2017) at 19-21; see a/so Arista Post­ Proceeding Submission (dated February 23, 2017) at 28 ("EntityManager is a library, which is a collection of code or tools that are used by Linux processes such as ProcMgr, Sysdb, or EOS agents. . . . EntityManager is a collection of code that 'can be used in general by [any process]."'). The incorporation of the EntityManager into the agent process is shown (and explicitly pointed out) in the picture Cisco incorporated into its Pre-Proceeding Submission (dated February 2, 2017):

[ ]

See Cisco Pre-Proceeding Submission (dated February 2, 2017) at 22. As Arista explains, "[t]he EntityManager library was called by running agents in the legacy EOS, and agents used EntityManager library code by importing that code into an agent's process memory or address space, and running the library code as part of the agent process itself." See Arista Post-Proceedi ng Submission at 29 (citations omitted). In contrast, agents "never call ProcMgr and never import ProcMgr code into the agents' processes or address space to run as part of a single subsystem" and "ProcMgr never calls agents or imports agent code into its processes generally. . . ." Id. The [ ] and the [ ], which Cisco claims should be combined with an agent process to form the "managing subsystem," are functions that are part of the ProcMgr process, they are neither run as part of the agent process nor are they part of a library (i.e., a collection of code that can be used in general by any process): the [ ] and the [ ] can only be called and used by the ProcMgr. This is unlike the EntityManager, which was run as part of

36 

the agent process and was a library or collection of code that could be used in general by any process.

Finally, Cisco raises some additional arguments relating to the “intertwined” relationship of the ProcMgr and agent such as the parent child relationship of the ProcMgr (parent process) and agent (child process). Cisco Pre-Proceeding Submission (dated February 2, 2017) at 17-18. This argument also is an insufficient basis for combining the ProcMgr because as Arista points out “ProcMgr used the fork() and exec() functions to start not only agents but also the Sysdb process” which means that “Sysdb starts as ProcMgr’s child process.” See Arista Pre-Proceeding Submission (dated February 2, 2017) at 31. Hence, defining subsystems based on parent child relationships “would lead to absurd results.” Id. Moreover, as Arista further points out, “all processes in Linux except for the very first (init) are started using the fork/exec method by which one process is started by another.” Id. at 32.

Alternatively, in the event that Cisco, as depicted in its Pre-Proceeding Submission (dated February 2, 2017), is arguing that the entire ProcMgr process should be combined with one specific child process to form the “managing subsystem” there likewise does not appear to be a basis for choosing to combine the ProcMgr with one child process, over the other child processes, to form the “managing subsystem.”

[ ]

Id. at 14. Further, while Cisco states that the ‘537 patent “does not exclude the use of common, shared processes[,]” as Arista notes, “combining ProcMgr or [ ] with a plurality of agents makes no sense” because there is only one ProcMgr and if that one ProcMgr were combined with all the agents to form a subsystem then that would effectively create only one client subsystem when the claims of the ‘537 patent explicitly state that there must be a plurality of client/router subsystems. See claims 1, 10, and 19 of the ‘537 patent; Cisco Pre-Proceeding Submission (dated February 2, 2017) at 27; see also Arista Letter (dated September 19, 2016) at 8. Moreover, in supporting the argument that “[t]he use of common, shared code to implement functions is replete throughout the industry,” and not excluded by the ‘537 patent, Cisco cites to libraries as an example for common code, however, as discussed previously, library code is distinct

37 

and notably different than program code and processes and it likewise cannot form a plausible basis for combining ProcMgr with a plurality of agents. See Cisco Pre- Proceeding Submission (dated February 2, 2017) at 27.

As such, it is determined that there are not sufficient reasons for combining certain functionalities of ProcMgr (or alternatively, the entire ProcMgr) with an agent to form the claimed “managing subsystem.” As the “managing subsystem” must be able to both (1) transmit the management registration request, and (2) externally manage data, there determination here is that this limitation is not met. While the agent is able to externally manage data, it does not transmit any request to Sysdb, and, assuming arguendo, that ProcMgr transmits a communication that qualifies as a “management registration request” sent to Sysdb, it is not able to externally manage data. Accordingly, Arista’s EOS Redesign does not literally infringe claims 1, 10, and 19 of the ‘537 patent.

Furthermore, under the test established in this ruling for application of the doctrine of equivalents, Arista has made a prima facie showing that network devices running the redesign EOS should not be excluded from entry on this basis. In particular, Arista has established that the redesign EOS does not have a managing subsystem that transmits a management registration request to the centralized database subsystem and that this claim limitation would be vitiated if a process, other than a managing subsystem, could transmit a communication that qualifies as a management registration request in order to satisfy the claim limitation in question. See Power Integrations, Inc. v. Fairchild Semiconductor Int'l, Inc., 843 F.3d 1315, 1344 (Fed. Cir. 2016) (“Under claim vitiation, ‘if a court determines that a finding of infringement under the doctrine of equivalents would entirely vitiate a particular claimed element, then the court should rule that there is no infringement under the doctrine of equivalents.’”) (internal citation omitted).

F. The `592 and `145 (Private VLAN) Patents

1. No Allegation of Infringement of the PVLAN Patents

As an initial point, and as indicated above, CBP is aware that the Commission has instituted a formal enforcement proceeding pursuant to Commission Rule 210.75(b), 19 C.F.R. § 210.75(b), based on a complaint Cisco filed on August 26, 2016, to determine whether Arista is in violation of the Commission’s cease and desist order issued on June 23, 2016. See Commission Notice, 81 Fed. Reg. 68455 (October 4, 2016). Notably, Cisco has not alleged infringement of the ’592 and ’145 patents by the EOS Redesign in the enforcement complaint and has not sought to introduce this aspect of the redesign during the pendency of that proceeding. See Complainant Cisco Systems, Inc.’s Enforcement Complaint, Doc ID 589164 at 12, n. 1 (dated August 26, 2016) (“Discovery may demonstrate that Arista’s continuing marketing and sale of its products also violates the Limited Exclusion Order and Cease and Desist Order with respect to the ’592 and ’145 patents; Cisco reserves its right to amend this Complaint to add allegations of infringement relating to those patents.”). Moreover, Cisco has not

38 

indicated to CBP during the inter partes proceeding that it believe there is infringement of the `592 or `145 patents by network devices running the redesigned EOS.

In light of the above, and given that Cisco has identified networking devices that implement this redesign during an investigation or related proceeding at the Commission but has not accused this functionality of infringement, CBP will not refuse entry on the basis of this aspect of the redesign until the Commission, under its authority, finds infringement and a violation of the Commission’s remedial orders as to this aspect. See CBP HQ Ruling H242025 at 9 (dated June 24, 2013) (“CBP will not refuse entry on the basis of features that were present on the accused devices at the time of the ITC investigation but were not specifically accused and found to satisfy the limitations in the asserted patent claims identified in the exclusion order.”)

IV. HOLDING:

The Arista Redesign EOS at issue is not covered by claims 1, 2, 8-11, and 17-19 of the '537 patent; claims 6, 7, 20, and 21 of the '592 patent; or claims 5, 7, 45, and 46 of the '145 patent. Therefore, the switches in question are not within the scope of the ITC’s limited exclusion order in Investigation No. 337-TA-944, and may be entered for consumption into the United States.

This decision is limited to the specific facts set forth herein. If Arista manufactures or imports switches that differ from the embodiment described above, or if future importations vary from the facts stipulated to herein, this decision shall not be binding on CBP as provided for in 19 C.F.R. § 177.2(b)(1), (2), and (4), and § 177.9(b)(1) and (2).

Moreover, this ruling letter, if found to be in error or not in accord with the current views of CBP, may be modified or revoked by a separate interpretative ruling under Part 177 of the Customs Regulations. See 19 C.F.R. § 177.12. When CBP undertakes review to determine if an issued ruling letter is in error based on a request from a person outside of CBP, the burden to establish that the ruling letter is incorrect, either legally or factually, falls on the person seeking to have that ruling letter modified or revoked. See CBP HQ Ruling H242026 at 6 (dated June 24, 2013) (“CBP does not require a compelling reason to revoke a prior ruling, even if there has been reliance on the ruling in question. Instead, the standard for CBP modification or revocation requires establishment that the prior ruling was issued in error or that it is not in accord with the current views of the agency. See 19 C.F.R. Part 177.12(a); see also International Custom Products, 549 F. Supp. 2d [1384,] 1394-95 [(Ct. Int’l Trade, 2008)]. With this framework provided, the decision now turns to the question of whether [the person requesting revocation] has established that HQ H226615 was issued in error.”).

Modification or revocation of a CBP administrative ruling must be carried out in accordance with the notice procedures set forth in 19 C.F.R. § 177.12(b) or (c), except as otherwise provided for in § 177.12(d). Specifically, under § 177.12(b), CBP may modify or revoke an administrative ruling that has been in effect for less than 60

39 

calendar days by simply giving written notice of the modification or revocation to the person to whom the original ruling was issued. A ruling under Part 177 is immediately effective upon its issuance unless the ruling itself provides otherwise. See 19 C.F.R. § 177.9(a) (“Generally, a ruling letter is effective on the date it is issued and may be applied to all entries which are unliquidated, or other transactions with respect to which the Customs Service has not taken final action on that date.”); see also 19 C.F.R. § 177.10(e).

However, when contemplating the issuance of an administrative ruling that would modify or revoke a prior ruling that has been in effect for 60 or more calendar days, CBP must follow the notice and comment procedures of 19 U.S.C. § 1625(c)(1), as implemented by 19 C.F.R. § 177.12(b)(1)-(3). See Int'l Custom Prods. v. United States, 748 F.3d 1182, 1185 (Fed. Cir. 2014) (“Section 1625(c)(1) requires Customs to follow multiple procedural requirements when issuing an ‘interpretive ruling or decision’ that would ‘modify . . . or revoke a prior interpretive ruling or decision which has been in effect for at least 60 days.’ In particular, Customs must publish the proposed ruling or decision in the Customs Bulletin, provide a comment period of at least 30 days after such publication, and publish the final decision in the Customs Bulletin within 30 days after the close of the comment period.”) (internal citations omitted). The effective date of a modifying or revoking ruling is governed by 19 C.F.R. § 177.12(e).

Lastly, the publication and issuance requirements set forth in 19 C.F.R. § 177.12(b) and (c) do not apply when a CBP position is “modified, revoked or otherwise materially affected by operation of law or by publication pursuant to other legal authority or by other appropriate action taken by Customs in furtherance of an order, instruction or other policy decision of another governmental agency or entity pursuant to statutory or delegated authority.” 19 C.F.R. § 177.12(d)(1).

As noted in the revoked ruling and earlier in this determination, CBP is aware that the Commission has instituted a formal enforcement proceeding pursuant to Commission Rule 210.75(b), 19 C.F.R. § 210.75(b), based on a complaint Cisco filed on August 26, 2016, to determine whether Arista is in violation of the Commission’s cease and desist order issued on June 23, 2016, with respect to the ’537 patent. See Commission Notice, 81 Fed. Reg. 68455 (October 4, 2016). The enforcement complaint confirms that this proceeding involves Arista switches running the re- designed version 4.16.6M of its Extensible Operating System. See Complainant Cisco Systems, Inc.’s Enforcement Complaint at 12, Doc ID 589164 (dated August 26, 2016) (“Exemplary Covered Products include the Arista switches, including at least the 7010, 7048, 7050, 7060, 7150, 7250, 7260, 7280, 7300, 7320, and 7500 series models and/or Arista EOS, including at least version 4.16.6M and later.”).

As with any determination of the Commission under section 337, as amended, 19 U.S.C. § 1337, including any determination under Subpart I to Part 210 of Subchapter C in Chapter II from Title 19 of the Code of Federal Regulations, if this enforcement proceeding produces findings of fact or conclusions of law that conflict with this ruling letter or are otherwise contrary in any other way, the ruling letter will be modified or

40 

revoked accordingly by operation of law under 19 C.F.R. § 177.12(d)(1) and CBP will carry out the agency action mandated by that determination.

Sincerely,

Charles R. Steuart
Chief, Intellectual Property Rights Branch

41