/hanoidb

Erlang LSM BTree Storage

Primary LanguageErlangApache License 2.0Apache-2.0

HanoiDB Indexed Key/Value Storage

Build Status

HanoiDB implements an indexed, key/value storage engine. The primary index is a log-structured merge tree (LSM-BTree) implemented using "doubling sizes" persistent ordered sets of key/value pairs, similar is some regards to LevelDB. HanoiDB includes a visualizer which when used to watch a living database resembles the "Towers of Hanoi" puzzle game, which inspired the name of this database.

Features

  • Insert, Delete and Read all have worst case O(log2(N)) latency.
  • Incremental space reclaimation: The cost of evicting stale key/values is amortized into insertion
    • you don't need a separate eviction thread to keep memory use low
    • you don't need to schedule merges to happen at off-peak hours
  • Operations-friendly "append-only" storage
    • allows you to backup live system
    • crash-recovery is very fast and the logic is straight forward
    • all data subject to CRC32 checksums
    • data can be compressed on disk to save space
  • Efficient range queries
    • Riak secondary indexing
    • Fast key and bucket listing
  • Uses bloom filters to avoid unnecessary lookups on disk
  • Time-based expiry of data
    • configure the database to expire data older than n seconds
    • specify a lifetime in seconds for any particular key/value pair
  • Efficient resource utilization
    • doesn't store all keys in memory
    • uses a modest number of file descriptors proportional to the number of levels
    • I/O is generally balanced between random and sequential
    • low CPU overhead
  • ~2000 lines of pure Erlang code in src/*.erl

HanoiDB is developed by Trifork, a Riak expert solutions provider, and Basho Technologies, makers of Riak. HanoiDB can be used in Riak via the riak_kv_tower_backend repository.

Configuration options

Put these values in your app.config in the hanoidb section

 {hanoidb, [
          {data_root, "./data/hanoidb"},

          %% Enable/disable on-disk compression.
          %%
          {compress, none | gzip},

          %% Expire (automatically delete) entries after N seconds.
          %% When this value is 0 (zero), entries never expire.
          %%
          {expiry_secs, 0},

          %% Sync strategy `none' only syncs every time the
          %% nursery runs full, which is currently hard coded
          %% to be evert 256 inserts or deletes.
          %%
          %% Sync strategy `sync' will sync the nursery log
          %% for every insert or delete operation.
          %%
          {sync_strategy, none | sync | {seconds, N}},

          %% The page size is a minimum page size, when a page fills
          %% up to beyond this size, it is written to disk.
          %% Compression applies to such units of page size.
          %%
          {page_size, 8192},

          %% Read/write buffer sizes apply to merge processes.
          %% A merge process has two read buffers and a write
          %% buffer, and there is a merge process *per level* in
          %% the database.
          %%
          {write_buffer_size, 524288},  % 512kB
          {read_buffer_size, 524288},  % 512kB

          %% The merge strategy is one of `fast' or `predictable'.
          %% Both have same log2(N) worst case, but `fast' is
          %% sometimes faster; yielding latency fluctuations.
          %%
          {merge_strategy, fast | predictable},

          %% "Level0" files has 2^N KVs in it, defaulting to 1024.
          %% If the database is to contain very small KVs, this is
          %% likely too small, and will result in many unnecessary
          %% file operations.  (Subsequent levels double in size).
          {top_level, 10}  % 1024 Key/Values
         ]},

Contributors

  • Kresten Krab Thorup @krestenkrab
  • Greg Burd @gburd
  • Jesper Louis Andersen @jlouis
  • Steve Vinoski @vinoski
  • Erik Søe Sørensen, @eriksoe
  • Yamamoto Takashi @yamt
  • Joseph Wayne Norton @norton