2

I modified my RDS instance to "Multi AZ : Yes". My primary RDS instance is in us-west-1a and for multi-AZ the secondary zone is shown as us-west-1c. I wanted to verify if whatever changes I am making on my primary database are getting copied to the Multi-AZ standby database quickly.

But I am not able to understand what endpoint URL should I use to login into Multi-AZ database. I am thinking the end point URL would be different from primary. Could you please help me on this?

hjpotter92
  • 78,589
  • 36
  • 144
  • 183
sagun
  • 33
  • 1
  • 4

3 Answers3

3

You do not have access to the secondary RDS instance in a Multi-AZ configuration. You just need to trust that AWS is replicating data correctly. In a Multi-AZ configuration, RDS will write to both replicas syncronously. It will not return the write request until both replicas have written correctly.

To access a Multi-AZ instance, you issue your reads and writes to the single RDS endpoint. In case of an issue, AWS will modify the DNS entry for that endpoint to point to the secondary replica. So as long as you are using the endpoint DNS record, and not caching the IP address when accessing the RDS instance, the failover process should be transparent to you with only a minute or so of "downtime".

Matt Houser
  • 33,983
  • 6
  • 70
  • 88
0

take a look at https://aws.amazon.com/rds/details/multi-az/. You don't typically interact with the replica(s) of RDS resources directly; AFAIK ( I'm not an rds expert ) you can't do what you're describing. The idea is that RDS does that for you, automatically keeping a consistent replica in a different AZ, and providing to you a consistent DNS endpoint.

erik258
  • 14,701
  • 2
  • 25
  • 31
  • 2
    Except for Aurora, where the term "Multi-AZ" means something so different as to be almost unrelated, this answer is essentially correct. It isn't really disclosed how Multi-AZ works, othet than that it does not use native asynchronous replication. It is, however, possible to [force a failover](http://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/USER_RebootInstance.html). The endpoint hostname doesn't change in this case, but the DNS changes to point to the second instance. – Michael - sqlbot Dec 17 '16 at 01:12
  • Thanks for sharing the link for force failover. I was under impression that I will not be able to verify "failover" scenario as I can only delete or reboot the DB. But reboot itself can kick in failover was slept from my mind. Thanks again. – sagun Jan 20 '17 at 23:22
0

Although OP asks for "verify data is copied quickly", Google pointed me here to "verify a multi-AZ RDS deploy". I'll share what I found in hopes that it's halfway helpful.

In the RDS console, there is an option on reboot to Reboot from failover which doesn't appear on a standard deploy.

Standard deploys do not have this option, which was a small but satisfying indication that the multi-AZ was acting as expected.

enter image description here

Source (and generally a pretty good read)

Q: Can I initiate a “forced failover” for my Multi-AZ DB instance deployment?

Amazon RDS will automatically fail over without user intervention under a variety of failure conditions. In addition, Amazon RDS provides an option to initiate a failover when rebooting your instance. You can access this feature via the AWS Management Console or when using the RebootDBInstance API call.

Ben
  • 54,723
  • 49
  • 178
  • 224