Static-Dictionary OZIP Codec for Low-Latency Random Access

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing data compression codecs, such as ZLIB and LZO, are inefficient for high-performance, low-latency applications with random data access due to their complexity and unsuitability for hardware implementation, leading to increased processing overhead and costs, as well as inadequate support for highly random access workloads.

Innovation Solution

OZIP, a data compression codec that uses direct token encoding to a static dictionary, enabling low-cost and high-performance hardware implementations with parallel processing capabilities, and supports random access through embedded block mapping, avoiding costly conditional branching and dynamic dictionary rebuilding.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of substance

If common compression codecs (ZLIB, LZO) are used, then compression ratio is improved, but processing overhead and hardware complexity increase

Engineering Contradiction:
Improvedata sizeVSAvoidhardware complexity
Core Design Contradiction:
Loss of substanceVSDevice complexity

Solution Approach 1:

The patent segments the compression process into distinct phases: dictionary construction from training data, tokenization of input data using the static dictionary, and encoding tokens with arithmetic coding. This segmentation allows each component to be optimized independently for hardware implementation, reducing overall system complexity while maintaining compression effectiveness.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary action by constructing a static dictionary during an offline training phase using representative training data. This pre-computed dictionary is then reused during the actual compression process, eliminating the need for dynamic dictionary construction during real-time compression operations, thereby reducing processing overhead and hardware complexity.

Inventive Principle:
Principle #10Preliminary action

2Loss of substance

If common compression codecs (ZLIB, LZO) are used, then compression capability is improved, but latency increases

Engineering Contradiction:
Improvedata sizeVSAvoidprocessing latency
Core Design Contradiction:
Loss of substanceVSLoss of time

Solution Approach 1:

The static dictionary is constructed in advance during an offline training phase, so that during actual compression operations, the encoder can immediately tokenize input data without spending time on dynamic dictionary construction. This preliminary action significantly reduces processing latency for real-time compression tasks.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent extracts the time-consuming dictionary construction process from the real-time compression path and performs it separately during offline training. This separation allows the main compression pipeline to operate with minimal latency, using only the tokenization and arithmetic coding steps during active compression.

Inventive Principle:
Principle #2Taking out (Extraction)

3Adaptability or versatility

If dynamic dictionary construction is used, then compression adaptability is improved, but hardware implementation cost increases

Engineering Contradiction:
Improvecompression adaptabilityVSAvoidhardware implementation cost
Core Design Contradiction:
Adaptability or versatilityVSEase of manufacture

Solution Approach 1:

The dictionary is constructed in advance during an offline training phase using representative training data that captures the characteristics of the data to be compressed. This pre-computed static dictionary provides adaptability to different data types while keeping the hardware implementation simple, as no dynamic dictionary construction logic is needed during real-time compression.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent separates the adaptability function into two stages: offline dictionary construction (which provides adaptability to different data types) and online compression (which uses the pre-adapted dictionary). This segmentation allows adaptability to be achieved without requiring complex dynamic dictionary construction hardware during real-time operations.

Inventive Principle:
Principle #1Segmentation

4Device complexity

If sequential access design is used, then algorithm simplicity is improved, but random access capability deteriorates

Engineering Contradiction:
Improvealgorithm simplicityVSAvoidrandom access capability
Core Design Contradiction:
Device complexityVSEase of operation

Solution Approach 1:

The patent performs preliminary action by constructing a block index during the compression process that maps uncompressed block offsets to compressed data locations. This index is built in advance and stored with the compressed data, enabling efficient random access without requiring sequential processing during decompression operations.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3120266B1OZIP compression and decompression
Publication Date: 2022.09.07 ORACLE INT CORP
  • EP3120266B1 patent drawingFigure 1A
  • EP3120266B1 patent drawingFigure 1B
  • EP3120266B1 patent drawingFigure 2A

AI summary

A method, apparatus, and system for OZIP, a data compression and decompression codec, is provided. OZIP utilizes a fixed size static dictionary, which may be generated from a random sampling of input data to be compressed. Compression by direct token encoding to the static dictionary streamlines the encoding and avoids expensive conditional branching, facilitating hardware implementation and high parallelism. By bounding token definition sizes and static dictionary sizes to hardware architecture constraints such as word size or processor cache size, hardware implementation can be made fast and cost effective. For example, decompression may be accelerated by using SIMD instruction processor extensions. A highly granular block mapping in optional stored metadata allows compressed data to be accessed quickly at random, bypassing the processing overhead of dynamic dictionaries. Thus, OZIP can support low latency random data access for highly random workloads, such as for OLTP systems.