The Oracle Berkeley Database was used to create the first of the Subversion back ends.

If the instructions given in the Subversion book are followed carefully, especially the section Supporting Multiple Repository Access Methods then the BerkelyDB back end will work well.

It can be quite difficult to keep to the file ownership rules defined in the book so using BerkelyDB can be problematic. This has led to many people attacking BerkelyDB maintaining it is too unreliable for general use. In fact, BerkelyDB is very reliable, provided the rather strict rules of access are followed.

Sleepycat are working with the Subversion project to fix some of the problems Subversion users have experienced.