I'm experiencing an issue with the automated nightly snapshots for our 170GB MSSQL database on RDS. The snapshots are taking around 2 hours to complete, and we're using a db.t3.xlarge instance with 4 vCPUs, 3000 IOPS, and 125MBps storage throughput. Our database doesn't have a lot of transactions, so this duration seems excessive to me. Is this typical, or is there something wrong with our current setup? I'm relatively new to RDS, having previously managed databases on-premises.
1 Answer
It sounds like your snapshot times could be impacted by running out of storage or CPU burst credits. Make sure to check your performance metrics in CloudWatch to see if that's the case. If your instance is crucial, you might want to consider upgrading to a non-burstable instance or at least get familiar with how the credit system works. That could really help improve your snapshot times.
Thanks for the reply! Which specific metrics should I be checking to assess this? I noticed the DB metrics are pretty much flat since the snapshot is more of a storage activity.