Search code examples
javamapdb

MapDB batch import perfomance

Is it possible to improve mapdb's HTreeMap perfomance on huge datasets by means of adding blocks of records instead of adding them one-by-one? I use filesystem.

Solution

  • You could import data using in-memory store and than copy it to disk. Also make sure that cache size is large.

    MapDB has Data Pump for fast BTreeMap import, HashMap will be supported in MapDB 1.4 eventually.