PTAB

IPR2013-00321

Oracle Corp v. Clouding IP LLC

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Dynamic Distributed Data System and Method
  • Brief Description: The ’449 patent describes a method and system that uses a tree-structured distributed directory to support access to, and track movement of, copies of data blocks in a distributed storage network. The system is designed to maintain cache consistency and utilizes a special, non-deletable copy of data called a "repository of last resort" (RLR).

3. Grounds for Unpatentability

Ground 1: Obviousness over Borrill, Baylor, and Futurebus+ - Claims 1-2, 5-7, 16, 27, 29, and 34 are obvious over the 1994 Borrill Article in view of Baylor and Futurebus+.

  • Prior Art Relied Upon: Borrill (a 1994 IEEE article titled "Extending Snoopy Cache Consistency to General Interconnection Networks"), Baylor (Patent 5,893,922), and Futurebus+ (an ISO/IEC logical protocol specification).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that the combination of Borrill and Futurebus+ discloses a distributed system for maintaining cache coherency that includes a special data copy called an RLR, which functions as a default owner responsible for storing the last, non-deletable copy of data. Baylor was asserted to teach a method for improving system performance by migrating a special copy (a "home" node) based on monitoring access requests via a "consultation count" and migrating the home when a threshold is exceeded. Petitioner contended that applying Baylor's migration logic to the RLR of the Borrill/Futurebus+ system satisfies the limitations of the challenged claims, including monitoring requests to the RLR, deciding to migrate it, and performing the migration.
    • Motivation to Combine: A Person of Ordinary Skill in the Art (POSITA) would combine these references to solve the known problem of improving data access latency. A POSITA would have been motivated to apply the migration motivation and policy from Baylor to the RLR taught by Borrill and Futurebus+ to move the RLR closer to the nodes accessing it most frequently. Borrill’s existing tree-structured directory for tracking data ownership provided a clear and known mechanism that could be readily adapted to track the migration of the RLR.
    • Expectation of Success: Petitioner argued a POSITA would have had a high expectation of success because the combination was predictable and desirable. The modifiable tree-structured network of pointers in Borrill provided a straightforward framework for implementing the migration taught by Baylor.

Ground 2: Obviousness over Baylor and Herlihy - Claims 1, 2, and 5 are obvious over Baylor in view of Herlihy.

  • Prior Art Relied Upon: Baylor (Patent 5,893,922) and Herlihy (a 1998 article titled "The Arrow Distributed Directory Protocol").
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner asserted that Baylor discloses migrating a "home" node but does so inefficiently by requiring a "split-home" architecture where a static portion remains fixed. Herlihy was presented as teaching a more efficient method for tracking mobile objects using a distributed, tree-structured "arrow directory," where only nodes along the path of movement need to update their pointers. The combination results in a single, non-split migratable home that functions as a default owner, equivalent to the claimed RLR, which is tracked efficiently using Herlihy's protocol.
    • Motivation to Combine: A POSITA would have recognized the inefficiency of Baylor’s split-home migration and would have been motivated to substitute Herlihy's more efficient arrow directory protocol to solve the known problem of reducing migration overhead. The objective was to achieve the performance benefits of migration without the high costs associated with Baylor's specific implementation.
    • Expectation of Success: Petitioner contended the combination would have been successful because it involved applying a known, more efficient solution (Herlihy's directory) to a known problem (high migration cost in a system like Baylor's). This would predictably result in a unitary, migratable RLR.

Ground 3: Obviousness over Soundararajan and Borrill - Claims 1, 2, and 5 are obvious over Soundararajan in view of the 1994 Borrill Article.

  • Prior Art Relied Upon: Soundararajan (a 1999 publication on memory use in multiprocessors) and Borrill (the 1994 Borrill Article).
  • Core Argument for this Ground:
    • Prior Art Mapping: Soundararajan was argued to disclose a migratable "home" node that serves as both a backing store (default owner) and a centralized directory. However, its migration protocol is expensive as it requires all system nodes to update their records. Borrill, as in Ground 1, teaches an efficient tree-based directory where only nodes on the movement path are updated.
    • Motivation to Combine: A POSITA would combine Borrill's efficient tracking mechanism with Soundararajan's system to reduce the expense of migrating Soundararajan's "home" node. This modification would improve the performance and scalability of the Soundararajan system by adopting a superior, known technique for tracking mobile data.
    • Expectation of Success: A POSITA would have reasonably expected success. Using Borrill's distributed directory would relieve Soundararajan's home node of its centralized directory function, simplifying its role to that of a migratable default owner (functionally equivalent to the claimed RLR), which could then be tracked efficiently.
  • Additional Grounds: Petitioner asserted an additional obviousness challenge for claim 28 based on the combination of Borrill, Futurebus+, Baylor, and Raymond (a 1989 article on a tree-based algorithm for distributed mutual exclusion). Petitioner also asserted an additional ground for claims 1, 2, and 5 over Soundararajan in view of Herlihy. These grounds relied on similar theories of combining known techniques for data migration and directory management.

4. Key Claim Construction Positions

  • Petitioner adopted the Board’s constructions from a decision to institute in a related IPR (IPR2013-00089) for the following key terms:
    • Repository of Last Resort (RLR): Construed as "a facility for storing a last or only remaining data replica that may not be deleted."
    • Storage Object: Construed as "any consistent storage entity such as a file, a block or an extent which may need to be updated automatically."
    • Name Tag: Construed as "any identifier associated with any object including storage objects."

5. Relief Requested

  • Petitioner requested the institution of an inter partes review and cancellation of claims 1-2, 5-7, 16, 27-29, and 34 of the ’449 patent as unpatentable.