Hi everyone! Now that Azure's NextGen Managed Instances are generally available, we're considering upgrading from our current General Purpose Managed Instances. I've been looking into potential downtimes since the official documentation suggests planning for it but doesn't specify how long it might be. I'm anticipating a micro downtime of only a few minutes when the failover occurs, but the vague phrasing about planning has me wondering if we might face longer downtimes, possibly exceeding 5 minutes. We're migrating Managed Instances that currently have around 80 databases and about 2-3TB of consumed storage. Has anyone made the switch from standard GP Managed Instances to NextGen GP Managed Instances? If so, can you share any notable experiences regarding downtimes?
2 Answers
According to Microsoft's documentation, you can expect a transition phase rather than a straightforward downtime. For an environment with 80 databases totaling 2-3TB, the bulk data transfer might take anywhere from several hours to even a couple of days, depending on your network and disk speed. Also, don't forget to account for the time needed for pre-migration checks and post-migration validations. If you're planning to go with an online migration, be prepared for the cutover window, which could last up to 72 hours. Just build in some extra time since the migration service can have a maximum lifetime of up to seven days.
Great info! It sounds like planning it during the weekend when the load is lighter might be a good strategy.
I've gone through the upgrade for all our Managed Instances, and the final step of the failover only took a couple of minutes for each instance. It went pretty smoothly!
Hey, thanks! Could you share how long the overall process took for each Managed Instance, specifically regarding data transfer and seeding? Also, what was the total storage size of the instances you migrated?
Exactly! I don’t know what the other person was suggesting, but it really just felt like another resize operation. They can be a bit tedious, but the results were completely predictable!
Thanks for the insights! I’ve been reading the documentation as well, and I’m estimating around 9 to 15 hours for the migration process. They mentioned an up to 10-minute downtime at the end, too, as some databases might take longer to come online after seeding. I think we’ll try this out, likely starting early Saturday or even Friday evening. I’ll keep everyone posted on how it goes! Cheers!