Top 3 cloud migration mistakes: summer 2019 version

Summer means grilling, vacation, and screwing up your cloud migration projects. Here’s how to avoid failure

Top 3 cloud migration mistakes: summer 2019 version

If you’re in the process of migrating workloads to the cloud, chances are this is not your first rodeo. Most of those doing cloud migrations are on their second or third project. Just when you would think we’re becoming more experienced, we’re also seeing an increase in migration failures. Here are the top three problems I see and ways to avoid them:

First: no larger vision beyond the projects. We’re building clouds by using tactical microarchitectures that are typically focused around a short sprint and a few workloads. Thus, we reinvent our cloud architecture each time. Soon we’ll be in cloud complexity hell, with no common services, such as common security, governance, managing, and monitoring.

It’s easy enough to understand how to solve the problem, but hard to get IT to do so. There needs to be some common macroarchitectural efforts, including defining common services, and a shared understanding of what the cloud is going to be to the larger enterprise. For some reason, we don’t like to think about cloud computing longer term, and that will kill the value you’re able to get out of cloud.

Second: lacking devops as part of the transformation. If you’re moving to cloud, then likely you need to move to devops as well. It’s a version of the razor and razor blade model as far as I’m concerned, but many enterprises are still not making the connections. 

By now we know how to set up devops, and we know that it allows you to get the most out of the cloud, such as speed to deployment, agility, and testing. Unfortunately, many enterprises look at devops as something that can be bolted on at the end of the migrations. That won’t work. Indeed, I’m not sure how you do migrations without devops.

Third: not doing knowledge transfer. This is another people issue. Now teams work on very different projects, gain lots of experience on those projects, but then share nothing.

There needs to be a central repository to collect artifacts, code samples, and lessons learned. I’m finding that cloud teams do this poorly, if at all. The result is people make the same mistakes over and over again, and nothing improves.

It’s an easy fix but you might need to create some motivation. For instance, on past projects I gave out $50 gift cards for each artifact provided or documented lesson learned. Participation increased to the benefit of $1 million (estimated), and cost me $5,000 in gift cards. That’s a deal I’ll do any day.

Happy summer!