search

krareT / pub-task

Terark public developing
MIT License
3 stars 6 forks source link

terark-fsa: Write an optimized dynamic Patricia Trie #2

Closed rockeet closed 6 years ago

rockeet commented 6 years ago

We can not find an optimized dynamic Patricia Trie implementation, so we write it ourselfves.

  1. Integrate to terark-fsa architecture
  2. Small memory usage
    1. Use a single memory block, be dumpable
  3. Fast lookup
  4. Fast Iterator

This Patricia Trie will be use as rocksdb memtable.

rockeet commented 6 years ago

With ConcurrentLevel: MultiWriteMultiReade, OneWritePerSubTrie, OneWriteMultiReade, SingleThreadShared,SingleThreadStrict, ReadOnly.

rockeet commented 6 years ago

Done.