RPO with percona

Hi community!

AWS using RDS they can reach to get restore point time of database 5 minutes using their service, is this possible to set in percona? at least 30 min doing by self configuration ? what hardware must to be consider if is possible to try a similar setup not using the RDS service?

our actual scenario is ec2 m5.xlarge (4 cpu,16 gb ram)
ebs gp3 with 10000 iops
100 database around 5gb ~ 50gb with more than 2000-3000 tables each.

right now we are using master - slave setup and doing backup using xtrabackup with xbstream from slave to send xtrabackup files compress and crypted to other server.

our backup full is taking 2 hours and diffs less than ~1 hour, we want to reduce that.

thanks

Hello @baph0m3t ,

low RPO can be reached with point-in-time recovery. In other words, you just need to ensure that you store binary logs somewhere that you can replay.

Each layer here lowers your RPO:

  1. Full backup
  2. Incremental (can be skipped)
  3. Binary logs

Hi @Sergey_Pronin

thanks for quick response, so the only need to do is do backup full and just keep the binlogs without delete it?
and where is point-in-time is a comand or an option to add to xtrabackup for restore use a command point-to-restore ?

I can’t recommend backup strategy without having the full picture. It may vary on case by case basis.

As for commands, please have a look here: Point-in-time recovery - Percona XtraBackup

ok thanks so much… now we evaluate the new way to get this goal and make some tests.