Best Practices for Backing Up Multiple Kubernetes Namespaces with Velero

[boedy]

Hello Velero Community,

I am currently fine-tuning my disaster recovery strategy using Velero and am faced with a decision that I'm sure many of you have tackled: when backing up a Kubernetes cluster with multiple namespaces, is it more advantageous to create a single scheduled backup for all namespaces, or to manage multiple scheduled backups, each including a single namespace?

Here's my context:

• My cluster has several namespaces, some of which are more critical than others.
• The amount of data and the rate of change vary across these namespaces.
• I am looking for a balance between operational simplicity and recovery effectiveness.

What is the recommended approach here?

[shawn-hurley]

Hello,

This depends on the RPO and RTO that you have. If they significantly differ from namespace to namespace, I suggest grouping them based on RPO and RTO.

One thing to note is that the backups happen sequentially, with only one backup at a time. So if you have a scheduled backup kick-off, say, five backups at the same time, they will only happen sometimes.

I would also be curious what others think

[shubham-pampattiwar]

+1 to what @shawn-hurley is suggesting here. Additionally,

• I would identify and prioritize critical namespaces, these may warrant individual backup schedules.
• For critical namespaces the backup schedules can be more frequent as compared to the non-critical ones.
• You could plan and execute a backup strategy right now and re-evaluate it down the line later and make changes accordingly.
• Overall I would suggest a hybrid approach.