I believe tow several weeks ago. I discovered a google's free project that may store key value pairs rich in performance. However i your investment title. Could anybody let me know? or possess some other recommendations for me? I've been using BerkerlyDB, however i found BerkerlyDb is not quick enough for my program. However, berkerylyDB is simple to use because it seems like a java lib jar, which may be integraed with my program effortlessly. My program can also be designed in Java.

Two strong rivals within the DHT (Distributed Hash Table) 'market':

  • Cassandra (produced by Facebook, being used by Digg and Twitter)
  • HBase

Here is really a presentation about Cassandra. On slide 20 you will see some speed benchmarks- .12 ms / write
(Searching around for the entire presentation, including Eric Evans speaking)

Perhaps you should describe what features you'll need. Whether it does not have to be distributed (will it?) i quickly would use the H2 Database. For individuals who think "it cannot be fast since it is using SQL" please be aware that after using prepared statement, SQL parsing is just done once. Disclaimer: I am the primary author of H2.

Many answer appear to instantly assume requirement for distribution but that appears odd if question describes BDB.

Knowing that, beyond Redis and H2 (that are both good), there's also Tokyo Cabinet to think about, which appears to provide benefits over BDB. And something more more recent possibility is Krati.

I believe you saw Guava or Google collections.