"We abandoned them in favor of B-trees in OpenLDAP 2.2 because extensive benchmarking showed that BDB's B-trees were faster than its hash implementation at very large data sizes."
Didn't bdb's linear hashing scheme extend the size of the hash table enough to keep it at the required loadfactor?
Thanks for the reply. Interesting that linear hashing had such a big effect, seeing as it is meant to be a slowly-occurring process that only happens when the average load factor of all buckets exceeds a threshold. I guess that was back in 2004 though. Wonder if the same performance is still applicable?
Didn't bdb's linear hashing scheme extend the size of the hash table enough to keep it at the required loadfactor?