Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

        Improve the availability with multiple replicas

        Better memory usage control mechanism

Performance

       2C8G 1M 128 dim Vector, serial search latency < 5ms

       2C8G 1M 128 dim Vector, 50 thread concurrent search latency < 50ms(QPS > 1000)

       Release vector search benchmark and tool 

       Support large cluster with tens of billion vectors and scale linearly

       Improve the hybrid search/retrieve performance

Major functionalities 

        Partition refine

...

       Change Data capture

       GPU Index support (Optional)

SDK

          Go SDK fully tested

          Restful API Documentation

          Refine python, java, go, node SDK and sync up all APIs

...

  1. Refactoring Data storage format to support new scalar execution engine and better compression
  2. Refactoring data coordinator and index coordinator, combine to single process and make it easy to understand and deploy
  3. Refine Implement QueryNode V2 to simplify search path and improve search performance 

...

  1. Support Huggingface , PaddlePaddle and OpenAI as the inference engine 
  2. Support Data Migration from Milvus 1.0, Faiss, Hnswlib to Milvus 2.0
  3. Support to deploy Milvus on GCP easily
  4. Feder support - Vector data visualization
  5. More trouble solving tools such as tracing, birdwatcher and dynamic config support

Documents

      1. Update all user document 

      2. First version of milvus technical guide