Atlas backup to second region
We would like the option to have our Atlas backups automatically copied to a 2nd region to support our DR initiative. I realize we could create a live database node in a 2nd region, but that is more costly and complex. Having the backup sync to another region that we can restore off of in case the primary region fails would be simpler and more cost effective.
Is this something that is on your roadmap, and if so, when will it be released?
Thanks!
Kyle
shared this idea
13 comments
-
D'Arcy
commented
This is definitely needed and I'm a bit surprised it doesn't exist already!
-
Sean
commented
This is a much needed feature for DR. Are there any work around anyone is using.
I am thinking about scripts to download backups and push them to S3.
-
Mike
commented
I would want to be able to independently configure the scheduling of this; e.g. so we could schedule the 2nd-region copies to be much less frequent in order to avoid too much egress bandwidth.
-
Geoffrey
commented
We need this functionnality in case of DR. My MongoDB solution architecte recommande us to vote here for this function.
Like asked by Kyle. Is it in the roadmap for cluster on Azure?
-
sean
commented
+1
-
Roger
commented
This would be great!!!
-
Fernando
commented
+1
-
Johnnes
commented
That would be really helpfull.
-
Vijay Katta
commented
+1
-
Chris Baird
commented
We fully support this same requirement. Having the continuous backups and/or snapshots in a second region can protect us against some very real logical failure scenarios. Having to replicate every microservice database to a live database in a second region does not seem to be a cost effective solution to this problem.
-
Xavier
commented
I agree with both comments. Having multi-regional snapshots in GCP would be very beneficial to us as well for DR purposes, and avoid performance degradation of geo-clustering.
-
Christopher
commented
+1 for this feature.
-
Hanspeter
commented
Some companies have a strong requirement to mirror the backup to a second region even if a cluster is stretched. So fully agree that this feature should be made available.