Hi everyone,
this is the first post I write here in the percona forum!
I’ve read a lot of blog post and watched a lot of webinars about percona xtradb cluster, but there’s something it’s not clear to me.
Scenario:
- 2 kubernetes clusters - 500Km each other apart on a private network
- On both of them a single application which writes on a Mysql DB. No Updates, only Insert and Select. 1 Insert per second on average, Select of the inserted data after one second
on average.
Goal:
Geo distribute the application. Share DB in Multi-Master sync replica.
I plan to use the percona kubernetes operator to deploy a percona xtradb cluster on each cluster.
I’ve got only 2 Datacenters on two different locations. So I plan to use an arbitrator to handle the even number of geo sites.
Questions:
-
Does it make sense to put an arbitrator on both 2 datacenters near each percona cluster?
I would have 2 arbitrators, but it’s not clear to me if this design could work. Otherwise, I should find another geographical site where to put the arbitrator. -
Feasibility: Is it enough test the RTT between the geo sites, and follow the “callaghan law” of 1 update per RTT?
Thank you so much.
Br
Syphon