Francesc Alted - Btune: Making Compression Better | PyData Global 2023

Discover how Bitune, a tool using genetic algorithms, optimizes compression parameters for improved compression speed and ratio, and learn how to balance tradeoffs for your specific use case.

Key takeaways
  • Bitune: a tool for finding optimal compression parameters based on genetic algorithms.
  • Compression time: can be optimized by using chunk-based compression and shuffling data.
  • Compression ratio: can be improved by using high-level compression algorithms like Blur Scale Z.
  • Neural network models: can be used for automatic guessing of best compression parameters.
  • Tradeoff: balancing compression speed and compression ratio can be achieved using the balance mode in Bitune.
  • Platform dependence: compression performance can vary greatly depending on the machine and operating system used.
  • Memory usage: compressing data in memory can be faster and more efficient than compressing data on disk.
  • Variable compression: using different compression algorithms and parameters for different datasets or use cases.
  • Genetic algorithm: can be used to fine-tune compression parameters for optimal performance.
  • Cache efficiency: compressing data into chunks that fit in caches can improve compression speed.
  • Z-standard: a high-level compression algorithm that achieves high compression ratios at the cost of speed.
  • LZ4: a fast compression algorithm that is suitable for real-time compression applications.
  • Bit shuffle: a filtering technique that can be used to improve compression ratios.
  • Chunk-based compression: compressing data in chunks can improve compression speed and reduce memory usage.
  • Split mode: can be used to split data into smaller chunks for improved compression speed and reduced memory usage.
  • Inference: can be used to predict the best compression parameters for a given dataset.
  • Performance mode: can be set to either compression or decompression to optimize for speed or compression ratio.
  • Environmental variables: can be used to set compression parameters and modes in Bitune.