I have an 2 node, AlwaysOn deployment on Windows 2012 R2 and SQL Server 2012 SP2. Both replicas are on 512b sector size drives.
We are migrating to a new VPC in AWS. In doing this we have added a replica to the availability group that is in AWS. This replica was created using SSD volumes in AWS, which are reporting themselves as 512e (512b sector size / 4096 physical sector size).
Now that this 3rd, AWS replica is in the availability group, I'm getting errors in my SQL log like:
There have been X misaligned log IOs which required falling back to synchronous IO.
I gather this is because the primary replica's log file is on a 512 byte sector size drive and the new replica's log file is on the 512e (512/4096) drive.
I have two questions about this I can't seem to find an answer to:
(1) Once we move completely over to the new VPC, and all drives are 512e, will this error message stop?
(2) Is there a performance issue having SQL server on 512e (Advanced Format) drives?
Any insight would be appreciated.
As a side note:
If I create the partitions on the AWS replica using Disk Manager the resulting volumes report 512b sector / 512b physical sector. If I create the partitions using Storage Pools the resulting partitions report 512b sector / 4096 physical sector.
I read that *all* SSDs have a physical 4k sector, so I'm assuming Disk Manager is simply creating the partitions unaware of the 512/4096 translation and the Storage Pools are aware of this; either way I think the translation is there.