When the conventional ways of doing things are ineffective, innovation is born

Levyx has re-thought traditional data processing and developed proprietary new methodologies for accomplishing more with the same hardware as before. For example, our data access algorithms are based on our own proprietary data structures. Traditional schemes are sub-optimal and needed updating.

Ultra high performance key/value storage

  • 1

    Designed to use flash and index billions of objects on a single node

  • 2

    High-level functionality: range-query, search and transactions

  • 3

    Pluggable database engine to "supercharge" database platforms and other applications and other applications such as Apache Spark, Kafka, Samza, Blockchain, and IOT

  • 4

    4. Wrappers support/embed popular Key Value Store APIs for Memcached, LevelDB and RocksDB

  • 5

    5. Integrated into major NoSQL and caching solutions and can be easily integrated into NewSQL, graphDBs, search platforms, streaming platforms

  • 6

    6. Software integration is accomplished using C, C++, Java, Node.js, and REST

  • 7

    Portable Implementation with Architecture and OS-specific Dependencies Fully Isolated
    - Available on Unix / Linux / Window / Mac platforms

  • 8

    Distributed in the Form of a Library
    - Fully documented key/value API

  • 9

    Platform for Integration with Other Technologies
    - Support for structured data (to improve Spark’s shuffle performance)
    - Columnar database integration with SparkSQL, Pentaho

I/O Performance

Performance Benchmarks

  • All values are in millions of operations per second
  • For each data point, a total of 1M operations per thread were performed

Helium Characteristics