There are many reasons companies consider going multi-cloud: to lift and shift their data center into the cloud, extend their data center into the cloud or to move applications to a different environment. Or maybe it’s because a single cloud service provider can’t meet all of their complex compute and storage needs.
Start with the right questions
Before you start multiplying your clouds though, it makes sense to take a step back.
“Far too many times IT managers and directors assume they can cram their entire infrastructure into one platform and that might not be the best case scenario for every project,” says Morgan Martensen, business development at IT hosting company SingleHop on the company’s tech chat, SingleHoppy Hour. “Our number one rule is: application first, infrastructure second.” That means asking questions about your applications before starting to plan what will power them.
Too many people don’t consider what’s driving the move. “It’s like walking into a car dealership and asking: ‘Do you sell cars?’ You have to have to understand what that IT director is trying to accomplish because there are a lot of delivery models,” says TJ Waldorf, VP of inside sales and marketing at SingleHop.
Asking questions is key, some to start with include:
- What workloads do they run
- What apps are they running and supporting
- What are the compliance requirements
- What are the geographical requirements
Keep an eye on the bills
Cost is often a key driver for going multi-cloud, but it may not be the simplest route. Instead of hashing out cost optimization on one platform, you’ve got five. “It’s definitely challenging,” says Waldorf. “If you’re using multiple vendors you’ve got multiple bills. AWS in particular…if you’ve ever seen an invoice from AWS it’s probably 47 pages with 10,000 line items that you’re not entirely sure what they are and when costs go up go or down it’s really hard to pinpoint why.” There are plenty of tools to help with cost optimization, but you still have to have the expertise to know how to run those tools and understand how to do it right, he adds.
Waldorf cites the case of one customer who moved from their own dedicated servers to the cloud looking for greater scalability, but realized that the unexpected bandwidth costs were much higher than anticipated. In a lot of cases, it makes sense to leave a few applications and workloads on dedicated servers or a private cloud. “It boils down to figuring out where those things need to go and optimizing for performance but also cost. There’s a lot of trade-offs with public cloud in any cloud platform,” he says.
Catch the whole 19-minute conversation on YouTube
- Digital Sovereignty – Why Open Infrastructure Matters - December 18, 2020
- OpenStack in Production and Integration with Ceph: A European Weather Cloud User Story - December 2, 2020
- #OpenInfraSummit Track: Public Cloud - October 12, 2020