Like their counterparts in other sectors, health IT departments are increasingly turning to multi-cloud deployments in order to, among other things, reduce downtime risks and increasing system-wide resiliency. But while these and other benefits of multi-cloud deployment are gaining popularity, they can run up costs quickly if managers haven’t planned appropriately.
As tech writer Kathleen Casey puts one issue, for example, “a multi-cloud architecture presents numerous application integration challenges, and poor workflow planning can drive up your bill. Cloud applications have many moving pieces, and in a multi-cloud model, it's possible that some app components will reside on different providers' platforms, as well as on premises.”
In her view, the first key to managing this kind of complexity is to create policies to determine the best platform on which to host an application component. Moreover, she says “one way to avoid high multi-cloud costs is to reduce the number of times an application component crosses a cloud platform boundary. . . . This makes your data center a central hub of sorts and ensures all workloads only cross a single cloud boundary; it's a similar model to the use of a single cloud provider.”
Similarly, Casey points out that most cloud providers “charge you when traffic moves in and out of their platform, and those charges can increase greatly in a multi-cloud architecture. When you use multiple cloud platforms, it's likely that data will cross provider boundaries, and every time it does, you pay more in network traffic costs. To reduce these network costs, review how your providers charge for traffic. Then, evaluate your workflows.”
Yet another way to manage multi-cloud deployments efficiently is to “make sure you know your requirements for all the services you want in a multi-cloud architecture and how much those services cost before you start any negotiations. But don't make a decision purely based on cost. Sometimes, a more expensive service brings more value, and you might choose one vendor for its big data capabilities and another for its transactional processing features.”
In short, while multi-cloud deployments can be effective tools, Casey says, “these tools aren't valuable unless you have a well-defined plan and cost optimization strategy in place.”