According to a recent survey from Virtana research, 82% of respondents now leverage a multicloud strategy. More than three-quarters (78%) deploy workloads on more than three public cloud providers. A full 59% run more than half of their workloads on a public cloud that’s part of a multicloud deployment. Finally, the survey found that 51% also plan to increase the number of public cloud instances they support in 2022. Even more interesting, 34% plan to use five or more cloud platforms.
We’ve monitored the multicloud inflection point for more than four years, so I don’t think many of these survey results surprise anyone. What I found interesting is that 63% of respondents reported that their organization relies on at least five separate tools for migration, cloud cost optimization, integrated performance monitoring, application performance management, and cloud infrastructure monitoring. A large percentage (83%) say they manually consolidate data from all these tools, perhaps using separate databases or even spreadsheets. Many operate these tools in isolation from one another. Only 17% claim to have automated integration of their tool data.
I’m not a big fan of spouting data from research companies for more than a single sentence in an article, but I thought the data here was interesting enough to warrant an exception. First, it proves the fact that multicloud complexity is a real problem. Second, it reinforces the need to pay special attention to the introduction of migration complexity before, during, and after multicloud deployments. Finally, it illustrates the lack of operational tool integration. Few operational complexity issues get solved if we deploy tools into silos; they will just generate more cost and risk.
Here’s my point as I sound yet another multicloud complexity alarm: You can avoid unnecessary complexity and ensure success with some additional planning prior to a multicloud deployment.
Consider these three points:
You don’t deal with architectural complexity by making things more complex. If you choose specific tools that only work with a single public cloud provider, complexity gets worse. If you cannot find a cross-cloud solution for common operational requirements such as security management, performance observability, process orchestration, and data monitoring and management, just to name a few, hold off deployment until you can find a cross-cloud solution to solve these problems.
When companies claim that “there are no cross-cloud solutions” for some operational processes, my experience is that they’re not looking in the right places, or they may have some bias against tools and technology they consider “off-brand.” Part of the solution is to open your mind to new approaches that might be a bit scary but that work.
You can’t make systems work better in a multicloud if they are already an architectural mess. Garbage in, garbage out—the old adage still applies. If you have poorly designed systems in the enterprise data center, don’t expect a miracle to occur when you move them to a public cloud. You’ll have to fix the problems before or during migration or your operational problems will be the same or worse.
The most important point is to design and plan how to operate your multicloud in the abstract and then develop a logical solution that addresses the operational requirements for all systems that will exist in the multicloud. This typically means that legacy systems, private clouds, edge computing, and several public clouds will be in the mix. Next, use that framework to pick the best enabling technology to automate most multicloud operations.
Understand that tools evolve over time. Some will be removed or replaced. That makes it doubly important to define the need that the tool will meet within the scope of our logical operations architecture rather than focusing on the tool itself. Those who select tools first fail most often. Or worse, they move forward with a suboptimized solution that ends up bleeding the company dry in avoidable operational costs.
A thorough plan is key to the success of any project. To succeed with multicloud, the project plan will require a lot of working, planning, and thinking. If a multicloud deployment is on your horizon, it’s time to get started on that plan. If you already have multicloud deployments in your rearview mirror, it’s time to run some audits.