Hi guys, This email is to keep you posted about the news on the development of OrientDB.
Starting from v2.2.4 we introduced *parallel transaction*. Before v2.2.4, transactions acquire an exclusive lock on the storage, so no matter if you have 1 or 100 cores, the execution was serialized. Now transactions are executed in parallel only if they involve different clusters (files). So, having 16 cores, you could set ALTER DATABASE MINIMUMCLUSTERS 16, so all the classes that you create after this command will have 16 clusters (files) per class. Now, if you have 16 threads, or 16 clients (if you are working with remote protocol and any non Java/Scala driver) you could bind the thread/client to a cluster id. Example: Client 1: BEGIN LET v1 = CREATE VERTEX V CLUSTER v_1 SET name = 'Luca' LET v2 = CREATE VERTEX V CLUSTER v_1 SET name = 'Austin' CREATE EDGE e CLUSTER e_1 FROM $v1 TO $v2 COMMIT RETRY 10 Client 2: BEGIN LET v1 = CREATE VERTEX V CLUSTER v_2 SET name = 'Luca' LET v2 = CREATE VERTEX V CLUSTER v_2 SET name = 'Austin' CREATE EDGE e CLUSTER e_2 FROM $v1 TO $v2 COMMIT RETRY 10 In this case the 2 transactions go in parallel, because they work on different clusters: _1 for the 1st client and _2 for the 2nd. If you have experienced poor performance with transaction, it's time to sue this powerful enhancement! Best Regards, Luca Garulli Founder & CEO OrientDB LTD <http://orientdb.com/> -- --- You received this message because you are subscribed to the Google Groups "OrientDB" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
