Cloudflare Inc v. Sable IP LLC

1. Case Identification

• Case #: IPR2021-00909
• Patent #: 8,243,593
• Filed: May 7, 2021
• Petitioner(s): Cloudflare, Inc. and SonicWall Inc.
• Patent Owner(s): Sable Networks, Inc.
• Challenged Claims: 1-44

2. Patent Overview

• Title: Mechanism for Identifying and Penalizing Misbehaving Flows in a Network
• Brief Description: The ’593 patent describes a system for identifying and controlling network data flows by analyzing behavioral statistics. The system tracks metrics like byte count and flow rate to identify "misbehaving" flows (e.g., peer-to-peer traffic) and enforces penalties, such as an increased packet drop rate, on them.

3. Grounds for Unpatentability

Ground 1: Claims 1, 2, 4-7, 17, 18, 25-27, 37, and 38 are obvious over Yung

• Prior Art Relied Upon: Yung (Patent 7,664,048).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner argued that Yung taught all limitations of the challenged claims. Specifically, Yung described a "traffic monitoring device" that classifies data flows by heuristically comparing observed behavioral attributes (e.g., packet count, byte count, inter-flow timing) against known patterns. Petitioner asserted that Yung’s creation of a "control block object" for new flows upon processing the first packet met the limitation of "creating a flow block." They argued Yung’s system updated these statistics for each packet processed, independent of network congestion, thereby meeting the patent's core requirements for storing and updating payload-content-agnostic behavioral statistics regardless of congestion.
  • Motivation to Combine (for §103 grounds): This ground relied on a single reference. However, Petitioner argued it would have been obvious for a Person of Ordinary Skill in the Art (POSA) to implement Yung's traffic management functionality in a single router, as Yung disclosed its system could be integrated into various network devices, including "gateways" (which a POSA would understand to be routers) and "bandwidth management devices."
  • Expectation of Success (for §103 grounds): A POSA would have had a high expectation of success because prior-art routers already performed classification and bandwidth management, making the integration of Yung's enhanced techniques a predictable improvement.

Ground 2: Claims 9-13, 19-24, 29-33, and 39-44 are obvious over Yung and Copeland

• Prior Art Relied Upon: Yung (Patent 7,664,048) and Copeland (Patent 7,185,368).
• Core Argument for this Ground:
  • Prior Art Mapping: Petitioner alleged this combination rendered obvious claims requiring the computation of a "badness factor." While Yung provided the foundational system for tracking behavioral statistics, Copeland taught a method for calculating a "concern index" (CI) value for flows to identify suspicious activity. Petitioner equated Copeland's CI value with the ’593 patent's "badness factor," arguing it provided a quantitative indication of whether a flow was exhibiting undesirable behavior based on tracked statistics like packet count.
  • Motivation to Combine (for §103 grounds): A POSA would combine these references to enhance Yung’s system. Yung expressly noted that its pattern matching could incorporate "other factors." A POSA would have been motivated to incorporate Copeland’s CI calculation to better identify and control undesirable traffic like network attacks and probes, a goal consistent with Yung’s purpose of managing "unruly, bandwidth-intensive applications."
  • Expectation of Success (for §103 grounds): Success was expected because both references relied on analyzing flow-based statistics. Integrating Copeland’s CI calculation into Yung’s existing statistic-tracking and analysis framework would have been a minor and predictable software update for a POSA.

Ground 3: Claim 3 is obvious over Yung and Four-Steps Whitepaper

• Prior Art Relied Upon: Yung (Patent 7,664,048) and Four-Steps Whitepaper (a 2003 publication from Packeteer, Inc.).

• Core Argument for this Ground:

  • Prior Art Mapping: This ground focused on claim 3, an article of manufacture with a data structure containing specific fields. Petitioner argued that Yung's "control block object" met most field requirements but that the Four-Steps Whitepaper was needed to explicitly teach a field for statistics about "dropped and non-dropped packets." The whitepaper, which described Packeteer's PacketShaper® product, disclosed tracking metrics including "counts and percentages of... dropped... TCP packets."
  • Motivation to Combine (for §103 grounds): A POSA would combine the references because both related to bandwidth management devices from the same company, Packeteer (Yung is a Packeteer patent). Yung itself suggested its teachings could be integrated with PacketShaper® products. A POSA would have been motivated to add the dropped-packet metrics from the whitepaper into Yung's system to expand its classification capabilities and provide better reporting on the effectiveness of imposed rate controls.
  • Expectation of Success (for §103 grounds): A POSA would have expected success because the combination involved incorporating a well-known feature (tracking dropped packets) into Yung’s system, which already tracked numerous similar metrics.

• Additional Grounds: Petitioner asserted that claims 8, 14-16, 28, and 34-36 were obvious over Yung and Copeland in view of Ye (Patent 7,295,516). This combination added Ye's teaching of enforcing penalties only when a "congestion condition is encountered."

4. Key Claim Construction Positions

• Petitioner contended that ten means-plus-function limitations in the claims should be interpreted under 35 U.S.C. §112(6).
• Petitioner identified the corresponding structure in the ’593 patent for performing all the claimed functions as a "misbehaving flow manager (MFM 210)," which is implemented via: (1) "one or more processors on a line card 202 [that] execute one or more sets of instructions," or (2) "hardwired logic components (e.g.,... ASIC's on a line card 202)." This construction was critical to mapping the functions onto the processors and software modules disclosed in the prior art.

5. Arguments Regarding Discretionary Denial

• Petitioner argued against discretionary denial under Fintiv, asserting that the parallel district court litigations were in their earliest stages, with petitions filed just seven and nine weeks after service of the complaints.
• Key arguments included:
  • No trial date had been scheduled, and any future date would be uncertain due to the ongoing pandemic.
  • Investment in the district court cases was minimal, as pleadings were not yet closed and discovery had not begun.
  • Petitioner stipulated it would not pursue the same invalidity grounds in district court if the IPR was instituted, eliminating concerns of duplicative efforts and inconsistent outcomes.
  • The petition's clear showing of obviousness and the existence of multiple parallel lawsuits weighed strongly in favor of institution, promoting efficiency and uniform resolution by a highly qualified tribunal.

6. Relief Requested

• Petitioner requested institution of an inter partes review and cancellation of claims 1-44 of the ’593 patent as unpatentable.