Secondary backup in another region

[thild42]

User Story

As a customer I want my data to be backup in another region that the service is deployed in So that I have a disaster recovery option for my service

Acceptance criterias

• Works for all engine except Kafka and M3 Aggregator
• Primary backup is still stored locally to the service region
• Secondary backup is stored in another region
• Feature is activated by default for new and existing services
• Feature is free of charge

Note

This feature is already provided for MongoDB.

Vote and Feedback

Ask us anything here in the comments below, and vote with emojis ! 👍 to vote for this issue

Discuss on Discord

Feel free to discuss with us on https://discord.com/invite/PwPqWUpN8G

[matmicro]

What is the meaning of the "backup" ?

• Is it a realtime sync/async backup node of my database cluster ? This means that in case of disaster recovery, i do not loose any data.
• Or is it a nightly snapshot ? This means that in case of disaster recovery i loose at last 24 hours of data.

[thild42]

Good question.

The full backups, the daily/nightly ones, are stored locally and in the secondary region. The real-time backups, the WAL files, are written every 5 minutes only locally and are used for point in time recovery.

[thild42]

Removing the "MongoDB" label as MongoDB backups are already stored in a distant region.

[thild42]

Good news, this feature is now released in production. You should be able to see the different location of your backups from the Control Panel.

The first is the local backup, and the second one is the remote backup.

In this example the service is running in Gravelines, so the local backup is in Gravelines, and the remote backup in Strasbourg.

[biapar]

How to restore from the secondary region and how to choose another region where to restore? When will be available for the instances?

[thild42]

Currently the restoration to another region is not permitted from the Control Panel and will be available soon. However the restoration to another region is permitted by the API, but there is a bug being resolved as we speak.

[Piscenois]

Hi ! Does this issue cover the last line in the table https://help.ovhcloud.com/csm/en-public-cloud-databases-postgresql-concept-high-availability?id=kb_article_view&sysparm_article=KB0049421#comparative-table ?

@thild42, when you said

The full backups, the daily/nightly ones, are stored locally and in the secondary region. The real-time backups, the WAL files, are written every 5 minutes only locally and are used for point in time recovery.

and I understand the secondary backup to another region is once a day, I wonder how RPO can be 5 minutes ?

Don't you need to backup the WAL every 5 minutes to another region ?

Regards

[thild42]

Hey @Piscenois, really good question !

It seems I was wrong when I say that WAL files were only stored locally. Both backups and WAL/oplog/binlog are created locally and sync remotely. So no worries regarding the RPO

I am in touch with our partner Aiven to get some clarification regarding the sync between local and remote drive (probably some rclone shenanigans).

[biapar]

The function disappeared from the control panel. I don't see it anymore.

[thild42]

Hello, @biapar I just check in the Control Panel and everything is fine on my end. Could you please send me your Public Cloud project identifier by email at thibaud.le-douarin@ovhcloud.com ?

[thild42]

Hello again @Piscenois.

I manage to get some additional information regarding synchronization and RPO. I can confirm that both backups and WAL are synched remotely. The synchronization method differs from engine to engine, for PostgreSQL pghoard is used, for MySQL it is myhoard, etc.

[Piscenois]

Thank tou @thild42