PTAB

IPR2021-00855

SAP Se v. Teradata US Inc

Log In
Key Events
Petition
petition

1. Case Identification

2. Patent Overview

  • Title: Method for Determining the Computability of Data for an Active Multi-Dimensional Cache in a Relational Database Management System
  • Brief Description: The ’357 patent discloses systems and methods for an "active cache" in a database system. Unlike conventional caches that only reuse previously stored query results, the active cache can aggregate or otherwise compute data already stored in the cache to answer new, different queries, thereby reducing the need to access the slower backend database.

3. Grounds for Unpatentability

Ground 1: Obviousness over Deshpande ’98 and Shim - Claims 1-7, 26-33, and 52-57

  • Prior Art Relied Upon: Deshpande ’98 (a June 1998 publication titled “Caching Multidimensional Queries Using Chunks”) and Shim (a July 1999 publication titled “Dynamic Caching of Query Results for Decision Support Systems”).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Deshpande ’98, a publication by the inventors of the ’357 patent, taught the foundation of the claimed invention, including caching multidimensional data in "chunks" for on-line analytical processing (OLAP) systems. Deshpande ’98 acknowledged its limitation was that all aggregations were restricted to the backend database and explicitly identified "aggregating chunks in the cache to get a missing chunk" as the logical next step for "future work." Petitioner asserted that Shim disclosed this exact missing piece. Shim taught a dynamic cache manager that used a lattice structure to answer a query by aggregating data from finer-granularity data cubes already stored in the cache. Shim’s "Split algorithm" determined when a query could be answered from cached data by searching for an "immediate alive ancestor" in the lattice. If multiple candidate ancestors (i.e., multiple aggregation paths) were found, Shim taught selecting the one with the "least estimated cost," which was proportional to the number of rows to be scanned.
    • Motivation to Combine: Petitioner contended that a person of ordinary skill in the art (POSITA) reading Deshpande ’98 would be explicitly motivated to improve upon its system by implementing the in-cache aggregation it described as "future work." Shim provided a direct, advantageous solution for this known problem by disclosing a detailed method to perform in-cache aggregation to improve query response times. The combination was presented as a straightforward implementation of a known research goal using a readily available, published solution.
    • Expectation of Success: A POSITA would have a reasonable expectation of success because Shim provided a detailed implementation, including pseudocode for its Split algorithm, which could be applied to the chunk-based caching system of Deshpande ’98.

Ground 2: Obviousness over Albrecht - Claims 1-5, 26-31, and 52-55

  • Prior Art Relied Upon: Albrecht (a July 1998 publication titled “On-Line Analytical Processing in Distributed Data Warehouses”).
  • Core Argument for this Ground:
    • Prior Art Mapping: Petitioner argued that Albrecht alone rendered the challenged claims obvious. Albrecht disclosed an intelligent middleware layer for a distributed data warehouse that performed aggregate caching. The system used a "patchworking algorithm" to answer queries by finding and combining partial aggregates, called Multidimensional Objects (MOs), stored at various levels of granularity across different distributed sites. When a query was issued, a "broker" would broadcast a request to data mart servers, which would bid to contribute their locally available MOs. The query optimizer would then select the combination of MOs from the bidding sites that required the "least aggregation and transportation efforts" to form the final query result. This process inherently involved determining if a query could be answered by aggregating cached data (the MOs), selecting from a plurality of aggregation paths (the different combinations of MOs), and answering the query using the selected path.

4. Key Claim Construction Positions

  • Petitioner argued that no specific claim construction is required. For the purposes of the IPR, Petitioner adopted the Patent Owner's apparent position from related litigation that the claim terms should be given their plain and ordinary meaning.

5. Arguments Regarding Discretionary Denial

  • Petitioner argued that discretionary denial under Fintiv would be inappropriate. The petition was filed early in a co-pending district court case, where trial was scheduled for January 2023, well after a Final Written Decision (FWD) in the IPR would be due. Petitioner asserted that the parties had not yet invested significant resources in the litigation, as the claim construction process had not even begun.
  • Furthermore, Petitioner argued that institution was warranted because the primary prior art references, Shim and Albrecht, which teach the key limitation of in-cache aggregation, were never considered by the Examiner during the original prosecution of the ’357 patent.

6. Relief Requested

  • Petitioner requests institution of an inter partes review and cancellation of claims 1-7, 26-33, and 52-57 of the ’357 patent as unpatentable.