Five years ago, the old-school, traditional way of managing IT environments didn't really work in the cloud. The all-in-one way of deploying cloud services for your average corporate IT group did not translate well to the siloed IT teams. It required fundamental change for all of the operational folks to get it done. Fast forward to today: the cloud is really just a virtual IT datacenter, with much less IT overhead.
Storage, compute, access - done.
The enterprise's barrier to entry for the cloud has never been lower. First off, adopting cloud services doesn't involve the usual IT hiccups and headaches. You don't have to find virtual server space, let alone physical space. You don't have to provision it the way it's dictated by IT. Normally, you have to wait for a ticket to be approved if new equipment needs to be purchased. Then there's the amount of time it takes for that equipment to be delivered, racked, configured, etc.
With the cloud, all of those blockers are removed. It's a wide open field for doing whatever you want to do, on your terms. It's just that simple—until it isn't.
The speed of innovation, change, and possibility is actually where the cloud gets complicated. New technologies arrive and disappear at a dizzying rate. The ones you want to—or hope will—stick around, you have to learn how to be responsible for. The major cloud providers are all ready to provide you with a vast array of tools and services to help you monitor your environment. They offer so many that it's almost impossible for IT leaders to determine what will work best for their environment to serve their needs now and into the future.
Having spent years in IT leadership and overseeing multiple complex cloud deployments, I've found it comes down to some key areas to consider before jumping head-first into the cloud:
Fuzzy logic is not good–you have to be clear on why. Are you doing a switch because there are tangible business benefits? (Do you need to be faster to scale? Are you throwing away your tech stack to try crazy ideas?) First off, determine which apps should migrate. My recommendation is to choose low hanging fruit, the apps that your team is super comfortable with troubleshooting. The cloud comes with a learning curve, so save the super-hard, complex stuff for when you're team gets more comfortable.
Secondly, create a baseline for comparison. It's the only way to understand what value you are getting–or not getting.
A deep understanding of the facts around how your cloud provider is compensated is essential. This is a true "buyer beware" situation–most providers would be happy to see you buy an XL environment, even when you only need a medium. So, be sure to do your homework!
3. Monitor Usage
It's the goal of everyone in IT leadership to enable engineers to do whatever the heck they need to do to make the cool stuff the company can sell. They should be romping in the fields, unfettered by mundane things like what stuff costs. We really do want that to happen. At the same time, budget's still have to be managed.
That being said, it is super easy to make mistakes and forget what you have running in the cloud. We've all heard stories about when that first bill from a cloud provider arrived and it was 10x higher than we thought it would be. Sure, the entire engineering team turned up cool stuff and did cool things. But then they moved on to the next project and it never occurred to them to think, "Oh yeah, I'm not using that anymore, I should turn it off."
The cloud should make things better, but how do you know if it actually is? By benchmarking. Y'know, the thing you did before you started (see item # 1 above)? That is what you use here to show the true benefits of your deployment.
Use your baseline performance metrics, your current performance and the administrative effort to maintain. These three factors (with maybe some happy, no-downtime stats) will justify your plan.
This is all a lot harder to do than you might imagine, but the capabilities after you've moved to the cloud will be exponential and you'll be a superhero. Being a superhero is good.
I always say that, for engineers, the cloud is the wild wild west--the possibilities are wide open! In reality, it's the wild wild west with guardrails. So, let the wild Rumpus begin!