By now, everyone has heard of the cloud, many people have experience with the cloud, and some folks have still not touched it. One of the primary reasons we hear that folks don't want to touch to the cloud is COST. There's a broad misconception out there that moving your infrastructure or applications to the cloud is going to cost a fortune in ongoing monthly subscription costs. This statement is absolutely not true. To calculate the true costs of the cloud, one has to really dive deep into the potential savings, as well as what else you are getting for that monthly subscription that you may not even be accounting for. Let's take a look at what a surface-level investigation into the cost moving a simple public-facing web application from an on-premise environment to the cloud.
Your company has a publicly facing web application hosted in your company's on-premise server racks (you have a load balancer with redundancy set up between your New York and Chicago offices). The server racks are aging and the on-board software is old and about out of support. You've been tasked with determining the cost of moving this application to the cloud. This application has steady usage for the most part, but usage spikes heavily when you push out a new version. Initial research for cost of moving this application might include points like the following:
This will add up pretty quickly! And all this to replace an application that's working perfectly fine! It might seem like moving to the cloud is a waste of money. As we mentioned, though, the server racks are aging and not only will the hardware not run forever, but the software on those machines won't be supported forever, so keep that in mind as we move forward! Now, let's take a look (again surface-level) at buying new on-premise hardware and software licenses for this setup:
TOTAL: $12,500 (one time)
As you can see, with these (very rough) estimates, it will take about 25 months before you're ahead of the game by buying new hardware/software licenses instead of moving to the cloud. Not too bad, but 2 years isn't a short time, either. Not to mention, by the time those 2 years are up and you've recouped the raw costs, your systems will absolutely be outdated, and upgrades may already be on the horizon once again. Additionally, there are many "what-if" scenarios that could emerge. What if the hardware just dies? What if the software goes out of support and leaves itself open to security vulnerabilities? While those what-if scenarios are certainly alleviated by moving your application to the cloud, they are hard to quantify if you're trying to calculate true costs, and they are certainly not guaranteed. So let's focus on the TRUE costs of each scenario, taking into account the often overlooked side effects of each scenario.
There are several costs that go into both on-premise and cloud costs that are often overlooked. For example, take into account some of the costs that you may not realize you're spending on on-premise infrastructure for applications.
These things add up and are often taken for granted when doing an analysis of cloud costs. "We're always going to need these things", is not really going to be applicable anymore, as more and more applications and infrastructure move to the cloud. There are, however, additional costs that come with a move to the cloud that are also overlooked.
Luckily, things like electricity, cooling and security are all handled by Microsoft (or other cloud provider) for you! So those costs are gone. Your dedicated Operations staff may opt to move to the DevOps side, but you will likely need less staff over time as more apps are moved to the cloud. That said, we will not take that into account for this exercise.
So let's look at revised costs, including hidden costs, for on-premise resources needed for our application. Assume that this is THE mission critical application for the company, and thus requires a good chunk of the shared resources.
TOTAL: $49,600 (rough)
And the comparative cloud costs:
Overall, it will take 59 months before we break even for the on-premise solution. Granted, the shared costs for staff could vary dramatically, but this exercise makes it obvious that there is much more cost that goes into keeping your IT apps and infrastructure in-house than just the cost of hardware and software. 59 months is MUCH longer than the 25 we calculated before. Again, a lot can happen in that time. Software will be upgraded, hardware will get cheaper and better, and who knows what else. When deploying to the cloud, you just don't have to worry about any of this! In addition, we signed up for geo-redundant backup services, which means that we don't even have to worry about disasters in a single data center since our backups are spread out geographically! So if you happen to have a remote branch set up for the primary purpose of creating on-premise geo-redundancy, perhaps you can one day shut that branch down and realize even more savings! Finally, your downtime is going to minimized as the 3 big cloud hosting providers are always doing everything they can to achieve maximum uptime. How much money does that save? After all, time is money! If your systems are down, your company is undoubtedly losing money!
There are so many hidden costs to hosting all of your applications and infrastructure on-premise. We took VERY rough guesses at what those costs were by diving deeper into a hypothetical on-premise setup. All of these on-premise costs also go with unwritten headaches that you can't put a price on. When you move to the cloud, those headaches are taken on by Microsoft, Amazon or Google! The "true cost" of going about your day without those headaches is PRICELESS!