Teradata Operations, Inc. v. Realtime Data LLC

1. Case Identification

• Case #: UNASSIGNED
• Patent #: 9,054,728
• Filed: January 30, 2017
• Petitioner(s): Teradata Operations, Inc.
• Patent Owner(s): Realtime Data LLC d/b/a IXO
• Challenged Claims: 1-3, 9-10, 15, 20, 24

2. Patent Overview

• Title: Data Compression System with Content-Dependent and Content-Independent Modes
• Brief Description: The ’728 patent describes systems and methods for data compression that use a combination of content-dependent and content-independent techniques. The system analyzes a data block to identify its parameters or attributes, selecting a specific compression algorithm if the content is recognized, and applying a default compression method if it is not.

3. Grounds for Unpatentability

Ground 1: Obviousness over Franaszek, Hsu, and Sebastian - Claims 1-3, 9, 10, 15, 20, and 24 are obvious over Franaszek in view of Hsu, with an alternative combination including Sebastian.

• Prior Art Relied Upon: Franaszek (Patent 5,870,036), Hsu (a 1995 article titled Automatic Synthesis of Compression Techniques for Heterogeneous Files), and Sebastian (Patent 6,253,264).
• Core Argument for this Ground:

  • Prior Art Mapping: Petitioner argued that the combination of Franaszek and Hsu taught every limitation of the challenged claims. Franaszek was asserted to teach a base data compression system that selects a content-dependent encoder from a "Compression Method List" if a data block's "type" is identified via a descriptor field. If the type is unknown, Franaszek selects a single encoder from a "default list" of compression methods by testing samples, which constitutes a content-independent approach.

    The key limitation added during prosecution, which Petitioner contended was the basis for allowance, was "analyzing data within a data block... [which] excludes analyzing based solely on a descriptor." Petitioner argued that Hsu directly taught this element. Hsu described a system for compressing heterogeneous files by analyzing the data within each block, not just relying on a descriptor. Specifically, Hsu taught examining the first, middle, and last 512 bytes of a data block to determine its type and calculating "redundancy metrics" based on the internal data characteristics to select the best compression algorithm. This analysis of internal data, rather than just a header or descriptor, was precisely the feature allegedly missing from the prior art during prosecution. The dependent claims were argued to be obvious as they recited routine features also taught by Franaszek, such as receiving uncompressed data (claims 2-3), associating a data token with the compressed block (claims 9-10), and storing the output (claim 15).

  • Motivation to Combine (for §103 grounds): A person of ordinary skill in the art (POSITA) would combine Franaszek with Hsu to improve Franaszek's system for handling increasingly common heterogeneous files. Franaszek's reliance on a simple data type descriptor was less robust than Hsu's method of analyzing the actual content, which better accommodates data blocks with varying internal characteristics. A POSITA would have looked to Hsu's well-documented techniques for analyzing internal data patterns and redundancy to make Franaszek's encoder selection process more accurate and effective.

  • Expectation of Success (for §103 grounds): A POSITA would have had a high expectation of success because both Franaszek and Hsu operated in the same field of data compression and addressed the same fundamental problem of selecting an optimal compression algorithm based on data characteristics. Integrating Hsu's internal data analysis into Franaszek's framework was presented as a straightforward application of known techniques to achieve a predictable improvement.

  • Key Aspects: The core of the challenge relied on the assertion that the patent was allowed based on a mischaracterization of the prior art. Petitioner's argument centered on demonstrating that the allegedly novel feature—analyzing data within a block beyond just a descriptor—was explicitly taught by Hsu, a widely circulated 1995 publication. The alternative inclusion of Sebastian was presented as a fallback argument. If the Board were to find that Franaszek did not teach a "single data compression encoder," Sebastian was argued to cure this by disclosing a single "generic" filter used when a data type is unknown, which is a simple and obvious substitute for Franaszek's process of testing a default list.

4. Key Claim Construction Positions

• Petitioner argued for a specific construction of the phrase "the data block being included in one or more data blocks" (recited in dependent claims 2 and 3).
• Petitioner contended this phrase should be interpreted as "the data block is included among a group of one or more data blocks." This construction was proposed to align the claim language with the specification's disclosure, which describes a continuous stream of data blocks being input into the compression system, rather than one data block being physically contained within another. This interpretation was important for mapping the prior art, which similarly processes streams of sequential data blocks.

5. Relief Requested

• Petitioner requested the institution of an inter partes review and the cancellation of claims 1-3, 9, 10, 15, and 24 of Patent 9,054,728 as unpatentable.