# Percona Cluster Performance Question **URL:** \ **Category:** Percona XtraDB Cluster 5.x\ **Created:** [January 21, 2019, 10:47pm UTC](https://forums.percona.com/t/percona-cluster-performance-question/6808 "2019-01-21T22:47:02Z")\ **Posts on this page:** 2\ **Page:** 1
**Author:** ![mwarble](https://avatars.discourse-cdn.com/v4/letter/m/f6c823/32.png) [@mwarble](https://forums.percona.com/u/mwarble)\ **Post date:** [January 21, 2019, 10:47pm UTC](https://forums.percona.com/t/percona-cluster-performance-question/6808/1 "2019-01-21T22:47:02Z")
I’m currently using Percona Cluster for the HA aspect and have my application configured to use only a single node at any given time in a 3 node cluster. This essentially treats one node as a master and the other 2 nodes as slaves. If I were to use all 3 nodes in a normal multi-master configuration, is there any performance benefit to this from a write standpoint? Since a write to one node is written to the others as well, it would seem you are putting the same burden on the cluster regardless if you write to a single node or load balance to all 3. I understand there would be a read performance benefit, I just want to know if splitting the writing between the 3 nodes would also provide a write performance benefit? Thanks. ---
**Author:** ![krunalbauskar](https://avatars.discourse-cdn.com/v4/letter/k/439d5e/32.png) [@krunalbauskar](https://forums.percona.com/u/krunalbauskar)\ **Post date:** [January 22, 2019, 3:48am UTC](https://forums.percona.com/t/percona-cluster-performance-question/6808/2 "2019-01-22T03:48:19Z")
Hello, It all depends on how your writes are configured. If you continue to write/update the same tables from 3 different nodes, there are increased chances of conflicts (again depends on data) and so you may need to retry failed transaction. But say you plan to update different tables from 3 different nodes you may see an improved performance given you are getting consistent data on all 3 nodes. (Read as you pointed always gain from 3 nodes) * * * Important: Please use PXC-5.7 for improved performance. Starting 5.7.17 we have fixed major performance bottleneck in PXC. Latest PXC is 5.7.24 Regards, Krunal