# Design question -- which DB solution is best **URL:** \ **Category:** Other MySQL® Questions\ **Created:** [November 30, 2017, 1:38pm UTC](https://forums.percona.com/t/design-question-which-db-solution-is-best/6038 "2017-11-30T13:38:23Z")\ **Posts on this page:** 1\ **Page:** 1
**Author:** ![Bob565656](https://avatars.discourse-cdn.com/v4/letter/b/85e7bf/32.png) [@Bob565656](https://forums.percona.com/u/Bob565656)\ **Post date:** [November 30, 2017, 1:38pm UTC](https://forums.percona.com/t/design-question-which-db-solution-is-best/6038/1 "2017-11-30T13:38:23Z")
Hello, I currently have a 3-node MariaDB Galera cluster, servicing an application. We are looking to have this application active in four DCs. The following are the requirements of the new setup [LIST] [_]each site may be independently active [_]all DB data is to be available on every cluster at each DC [_]if replication (or sync) fails between one DC and another, both DCs shall continue to operate independently until replication is able to start up again (ie no split brain) [_]self sustained (no manual intervention required at any time) [_]ideally would like to have automatic collision handling [_]ideally would like an RDBMS solution (as we’re using jOOQ for Java) [/LIST] Things we can tolerate [LIST] [_]slave lag is acceptable [_]ACID not required [\*]as long as the data “syncs at some point”, it’s good enough [/LIST]Glossary 3NC = 3 node cluster DC = data center Research done so far: [LIST] [_]MYSQL group replication [LIST] [_]closest thing to what I need, however, it’s not recommended for WAN [_]only supports 9 nodes [/LIST] [_]MariaDB Galera (3NC at each DC, part of one global cluster) [LIST] [_]synchronous replication between 4 DCs is out of the question [_]if a DC drops out, it’s unusable [/LIST] [_]MariaDB Galera (independent 3NC at each DC, using bidirectional MYSQL replication between each DC) [LIST] [_]difficult to deal with auto increments and offsets, especially when adding new DCs or nodes [_]if there is any problem (like collision), replication fails and becomes a pain to start again [/LIST] [_]MYSQL NDB Cluster [LIST] [_]engine change from InnoDB, would rather not do that [_]downed data nodes may have to be brought up manually, the backup mechanism seems like a pain [_]on paper, geographical replication would work best for my requirements [/LIST] [_]Percona XtraDB Cluster [LIST] [\*]from what I can tell, it doesn’t support what I’m trying to do [/LIST] [/LIST] Do other SQL setups have a solution, maybe PostgreSQL?