0

I have an AWS ElastiCache instance running at around 85% memory capacity. Before scaling to a bigger instance, I would like to take a manual backup and test a few things out.

Failure of this node is not desirable as it has critical data and would cause downtime. Is there a possibility that the instance may die while taking a backup?

Redis version: 2.8.24

abhijeetpawar
  • 3
  • 2
  • 2
    Why is there critical data on a cache instance? There's a possibility the instance may die *at any time*, not just when taking a backup. – ceejayoz Jun 18 '18 at 21:02
  • @ceejayoz I agree that this isn't a good architecture. The plan is to move away from this. But it is what it is at the moment. Also, I'm aware that hypothetically the instance can fail at any time, but during backups, I would expect the backup process to fail and not the instance itself. – abhijeetpawar Jun 18 '18 at 21:09

1 Answers1

1

The best option here is likely to take a snapshot of the Redis instance using the build in features. This will degrade performance but shouldn't take it down. Note that it can fail at any time.

This seems like a scenario where a read replica would be useful, as it is automatically promoted to master if the master fails. You can add a read replica to a running instance. At that point you could take a backup of the read replica to do your backup.

The ElasticCache FAQ has useful information.

Tim
  • 30,383
  • 6
  • 47
  • 77