![[Update] Restoring Amazon Aurora DSQL multi-region clusters from backups is now easier](https://images.ctfassets.net/ct0aopd36mqt/wp-thumbnail-61c3c74c9fbde5b3bc234043f3ce3128/3c555f171fd799831bf6aeb586ca13bf/aws-backup)
[Update] Restoring Amazon Aurora DSQL multi-region clusters from backups is now easier
I'm Iwasa.
Amazon Aurora DSQL has the concept of single-region and multi-region clusters.
In a recent update, multi-region clusters using Tokyo, Seoul, and Osaka were supported.
For these multi-region clusters, recovery operations were actually more complicated compared to single-region clusters and required several steps, but with a recent update, restoring multi-region clusters has become much simpler.
First, previously, recovery points for multi-region clusters weren't supported for operations in the management console, but with this update, operations through the management console are now possible.
And previously, after performing recovery procedures in each region using AWS CLI, you needed to reconfigure the multi-region setup within 24 hours, but all of that is no longer necessary. It does everything automatically for you.
Let's try it out.## Trying recovery operations with source region: Tokyo, peer region: Osaka
The previous blog had Tokyo as the source region, Seoul as the peer region, and Seoul as the witness region.
We are restoring this multi-region cluster using an on-demand backup created from the Tokyo region cluster.
There is one important point: the same recovery point needs to be copied between regions, just like before the update.
Therefore, for this test, we sent an on-demand backup created from the Tokyo region cluster to the Osaka region using a copy job.
The operation has become very simple - from the AWS Backup console, select the recovery point in the Tokyo region and press the restore button.
It's the same as restoring a single-region cluster.
Then, in the restore options, you can configure the multi-region cluster settings, so select "Add peer region" here.
You can then specify the peer region and witness region.
The original multi-region cluster had Seoul as the peer region and Osaka as the witness region, but let's try specifying them in reverse.
After that, all you need to do is wait as the restore job handles everything nicely. In this case, AWS Backup restore jobs are running in both the Tokyo and Osaka regions.
The restoration has been completed.
My multi-region cluster was in an almost initial state with hardly any data created, but the restoration time was about 10 minutes.## If there is no identical recovery point in another region, a "No associated recovery point found" error occurs
If you specify a peer region without having copied the recovery point from the Tokyo region to the Osaka region, an error indicating that the recovery point cannot be found will occur as shown below.
Make sure to confirm that the recovery point has been copied to the peer region.
Copying to the witness region was not required.
In conclusion
Today, I tried the simplified process for restoring Amazon Aurora DSQL multi-region clusters from backups.
Although I had never performed a restoration operation before this update, as of today, you can check the information by switching to Japanese on the AWS Backup / Aurora DSQL multi-region cluster restoration pages.
Previously, the procedure was cumbersome because the management console wasn't supported, and you had to change the multi-region cluster configuration settings after restoring via API. With this update, the process has become much simpler.
Also, as tested in this blog post, this restoration method can be used to change one of the three multi-region configurations, such as changing the witness region.
In this case, we were able to change the witness region from Osaka to Seoul.
