
The main feature I found as a user and from the benchmarking perspective was that it is embeddable. I can simply use the database by including a jar and update it when necessary. There are no mash-ups or complications of having to explicitly keep the server started for the database. In addition to having implemented the Tinkerpop stack, which is the standard for graph APIs, there is little transition time between using other graph databases and OrientDB. And, of course, the transition from SQL to Graph becomes simple since it supports SQL-based syntax. Review collected by and hosted on G2.com.
Many times when the nodes go down (using a distributed configuration), there are inconsistent states that remain unresolved. The database has to be deployed again. Also, when I scaled my cluster, I had to manually move all the data from the smaller cluster to the larger one. Review collected by and hosted on G2.com.