During our recent SentryOne Accelerate 2020 virtual conference, I led a cloud migration panel discussion with several SQL Server experts. I asked the panelists to imagine they were riding in an elevator and someone turned to them and said, “You are really good at managing SQL Server. What tips do you have for someone like me who is going to undertake a cloud database migration soon?” With that in mind, I asked them to share their thoughts on how to be successful when planning for a cloud database migration.
Our esteemed cloud migration panelists included:
- John Sterrett, principal consultant and founder of Procure SQL LLC, Microsoft Certified Trainer (MCT), and Microsoft MCSE: Data Platform
- Dustin Dorsey, director of data management at LifePoint Health
- Erin Stellato, principal consultant at SQLskills and Microsoft Data Platform MVP
- Patrick Kelley, customer success engineering manager at SentryOne
Check out the excerpt from the discussion where the panelists shared the tips they would offer data professionals who are planning to migrate their databases to the cloud.
Cloud Database Migration Considerations
Kevin Kline: What tips would you share with data professionals who are preparing to move their databases to the cloud?
Patrick Kelley: You don’t know what you don’t know, so especially in these complex environments, having a solution like SentryOne Document, where you can really understand the breadth of “What are all of our data flows?”, “What’s communicating with what?”, or “If we’re going to move this, what are all the different ways that’s being leveraged outside our siloed scope so that we can prepare for moving forward?”
In terms of cost, when we know things are running poorly, we obviously want to tune those things. And that becomes even more true in the cloud because you’re paying more for poor performing queries or bad code. For on-premises databases, your costs are sunk at the time of purchase. But for the cloud, you pay for every CPU cycle, for every wasted IOP, and for unnecessarily consumed storage. Those costs add up quickly. Also, to think about that generally with all the stuff that data pros have to deal with day in and day out, it’s easy to say, “Hey, if it ain’t broke, don’t fix it,” right? But the reality is that just because no one’s banging on your door saying these things are running poorly does not mean there is no room from improvement.
With something like SentryOne and modern solutions aggregating that data and get that top 20%. “What’s our most heavy hitting, worst-performing 20% of our workload?” And if we can even chip away at a little of that, whether things are on fire or not, that is going to save you money in the long run.
John Sterrett: Testing. And that’s not only testing for performance. But I think cost as well to make sure that you’re actually going to be able to deliver what is expected at the price point and also with being able to keep everyone as happy as they currently are before you make a big move up to the cloud. You can get the performance you want from the cloud, but at what cost? Only testing will reveal that answer before you deploy.
Dustin Dorsey: A couple of things is spending the time upfront and establishing a good framework and good governance in the beginning. This is something that helps you immensely with cost management once you get down the road and start managing your cost. And make sure you have good structure and how you’re organizing your resources; having security in place of just having a good framework that you can build on.
To someone who’s new to the cloud, the best piece of advice I could give is to start slow. Don’t take your mission-critical application and make that your guinea pig you’re going to move over. You’re going to make some mistakes. You are going to bang your head a couple of times. Take something that is a light workload that can help you build some more familiarity and get used to it. Move that, see how it translates from on-prem to a cloud solution. Look at your costs and make sure the cost lines up with what you originally estimated. Make sure that you know that you set your expectations based off of that so that when you do move your mission-critical application it goes a little smoother and you know what you expect.
The last thing is to be prepared to automate and automate a lot because you are going to need an automation for sure.
Erin Stellato: I would echo all of that. I think this is important: You need to try to get support and buy in from everybody on the team, because if you do not, it’s a big challenge. It can be more complicated and take longer than it really needs to if people are dragging their feet.
My recommendation would be that when that decision is made, you sit down with all the key players—the DBA team, whoever else is going to be involved, the application team, etc.—and say, “What are your concerns? Why don’t you want to do this?” Give everybody the opportunity to speak their piece. If people have the ability to express their concerns and have their voice heard—even if you don’t end up doing what they would prefer to do or even if you end up migrating although they don’t want to—the fact that you’ve given them the chance to state their concerns and get their opinion out will help you down the road.
Then, for those folks who are concerned, for maybe your DBAs who do not feel confident in their Azure skills—support them. Get them training, provide them funding and encourage it, so that you are setting them up for success. I think that for so many of us, Azure, even though it’s SQL Server, just feels and looks so different. But once you understand all the underlying pieces, you start to become more confident. Build that into your people from the get-go.
Kevin Kline: Excellent advice. In fact, I endorse everything said so far. I feel that most of the time, we as technologists just expect our biggest challenge to be the technology itself. I have found after a long career it is really more of the people that cause us to have big issues or really big stumbling points. Especially if your team is not quite cohesive or everybody is not on the same wavelength. That is where you can see some big contentious fights happen.
Erin Stellato: Yeah, if you ask a business what their most important asset is, a lot of times they’ll say it’s their data. But the thing is, without your people, you can’t do anything with your data. You must pay attention to both.
Related Resources
Check out the following resources to learn more about the topics and tools shared by the cloud migration panelists.
- Webinar: Introduction to the SQL Server Database Experimentation Assistant (DEA), hosted by me and John Sterrett, which walks through how the DEA can help you save money, validate performance, and make sure you do not have errors while migrating to the cloud
- Blog: Identifying Cost-Saving Opportunities in Azure Dev Ops, which covers ways in which you can examine your spend in Azure DevOps, including auditing license usage and pipeline efficiency
- Blog: Building the People, Processes, and Technology for Data-Driven Culture with Intelligent DataOps, which discusses how a solid DataOps culture helps companies rein in data chaos, monetize data, actualize business, and drive cloud migration
- Cloud Migration Guide: SQL Server to AWS and Cloud Migration Guide: SQL Server to Azure, which provides a detailed look at the migration process and options and walks through how SentryOne solutions can assist with the upgrade and migration of SQL Server platforms to AWS and Azure
