Success with multicloud architecture is more about ongoing multicloud operations, not how optimized the design is.
Good multicloud architects are hard to find. Most of those calling themselves multicloud architects really just specialized in a single hyperscaler and found that things turned multicloud quickly, so they did as well. It’s called “multicloud washing.”
Creating the right multicloud solution, one that is nearly fully optimized, requires that you focus more on what’s between the public clouds, and not what’s in them. This requires a new level of understanding that most cloud architects don’t have yet, but that will change quickly during the next few years.
A few principles are emerging around configuring, building, and deploying a multicloud architecture. At the core is the missing focus on the resulting operations, which is where most multicloud designs get in trouble.
It’s one thing to define a multicloud, which is a complex array of technology, including common services such as security, governance, monitoring, data management, etc. It’s another to define how all of this is going to be operated longer term, including the resources and costs required. In many instances, the cost of operationalizing most multicloud configurations is too much relative to the value it’s able to bring back to the business.
There are many reasons for this, but it’s mostly complexity. Too many types of technology, brands, and approaches have led to a multicloud solution that’s overly complex and leads to operational approaches and resource requirements that are unrealistic.
I had one case where the cost of operationalizing a multicloud approach was five times that of the “as is” state. This wasn’t known until after deployment, and they had to scurry around for years normalizing technology within the multicloud to get to a more reasonable level of operations. Millions of dollars were lost directly plus the lost business opportunity of having more agile technology.
It’s time to change the way we approach, measure, and deploy multicloud. It’s common for multicloud architects to focus too much on the upfront design without regard for how the selection of technologies running in and between the cloud providers will be operated successfully long term. Moreover, the cost includes having different talent to support the number of technologies you’ll be employing, the tools, and the cost of risk. Overly complex multiclouds are more likely to be breached, considering that complexity leads to security mistakes.
What’s fundamental about this is that it’s just fundamentals. Architecture has always been about dealing with the long-tail aspects of the architecture, namely operations. For some reason, enterprises aren’t focusing on this, and when they move to multicloud they get into trouble. It’s preventable.
