Reputation: 9013
There are 3 options of Replication to deploy Azure project: locally-redundant, geo-redundant and RA geo-redundant. Microsoft says, that locally redundant creates 3 copies, geo-redundant - 6 copies (in 2 datacenters) for protection. And it costs in 2 times more. It's interesting for me - geo-redundant allows more faster access (because 2 datacenters in different regions) or not? Or it's so small, that it can be neglected? Also it's interesting about RA geo-redundant
Upvotes: 0
Views: 2057
Reputation: 1751
Geo-Redundancy is there to ensure failover for major disasters.
If you look at the scalability and performance targets you'll see that geo-redundant storage has lower throughput values (ingress and egress). You have to set it to read-access geo-redundant storage to expose the secondary endpoint (i.e doesn't use the same address as primary). You could factor this failover option into your apps/web services if you want, but with it being read-only it would still be stunted.
I guess it depends on a) how important your data is (always nice to tell your clients that their data is safe even in a disaster) and b) whether you'll realistically reach any of the limits of the storage accounts.
I haven't served media files of any size from the storage account so I tend to run into entity/partition limits before ingress/egress limits. I would enable geo-replication from the start and disable it if/when you hit performance issues (or split data between multiple storage accounts). There is a cost to re-enable geo-replication if it's been disabled, I assume this fee is due to having to a potentially large chunk of data to the secondary server.
my two cents
Upvotes: 2