At OpenArc, we work with companies that are building software solutions in the public cloud or moving existing workloads to the public cloud, and we’d like to share some of the things that we’ve learned. We pulled together two local experts and created this video to help you get started.
Here’s the video transcript if you prefer to read through it.
1. A number of our clients want to move to the cloud for cost savings. What are some good reasons for doing that?
Well, first of all there’s, obviously, a lot of hype around moving to the cloud and the savings you can have there, many analyst reports, lots of market-speak, but we’ve found that for some clients, moving existing workloads to the cloud can actually be more expensive than they expect it to be. Now that public cloud’s been around a while, there’s more evidence about what the real costs are. There are several reasons for this. If you take an existing workload and move it to public cloud, and you’ve got that workload currently sized to be the most massive load that you’re ever going to see on the system, then you’re going to be running the same thing you were running on-prem basically, so the costs are not going to be significantly different.
Another reason is that companies often don’t deprovision hardware when they move to public cloud, so they might end up in a situation where it’s a hybrid solution that some of it is running in public cloud, some of it’s running on-prem, or in their data centers and that’s the most expensive way to run a solution. There are often good reasons to do that, but it doesn’t always make sense. Moreover, if you don’t understand cloud pricing, each of the services that the different cloud providers provide is priced differently, and you need to understand how those things are priced to take advantage of those pricing models and get the cost savings that you’re looking for.
2. What are some specific examples of this?
When I did some work for a larger healthcare software company, we did some analysis to look at moving some of their very large software solutions into public cloud and with the cost models that we put together. We found that it was going to be actually 20 to 40 percent more expensive for them to run those workloads in the cloud. Now, again, this gets back to if you are going to take your existing workload that’s optimized for the highest load you’re ever going to see and move that into the cloud, you are going to pay money to do that, it’s not going to be cheap. We had to think about some different things that we could do to offset those costs and understand how to leverage the different services that were available in the public cloud to be able to realize cost savings in the long run.
3. Does that mean that there are no-cost savings in the cloud?
No, there definitely are but you have to think in terms of the total cost of ownership in the long run and not just consider lift-and-shift pricing or building new solutions, what it costs to price them, so you need to consider how to take advantage of the different services that the cloud provides. Some things to consider, one of them is, as we’ve already mentioned a couple of times, you don’t want to move your existing workloads precisely as they are today. You need to right-size them for the cloud before you move them up. Bain & Company did some analysis, and they found that companies that spend time right-sizing before moving to the cloud could save as much as 16% in their cloud migration costs, so obviously, that can be a big difference.
Some other things to consider are consumption pricing in the cloud. The platform as a service and serverless things that are available in these public cloud solutions, that’s where you can see real cost savings, so you need to think about how to transition architectures or how to build new things to take advantage of consumption pricing. If you do need to have always-on availability, then you might look at something like Reserve Instances, which is basically signing longer-term contracts with cloud providers so that they will give you discounts for that long-term commitment. This is another way you can offset, but an important thing to consider in public cloud is also taking advantage of the services that you’re paying for.
It’s not just about running a VM, you also need to think about the high availability that they make available to you, the ability to spin up and tear down different environments quickly, and things like setting up disaster recovery solutions that you don’t have to keep on all the time, you can actually spin them up when they become necessary. These are all of the different things that you need to understand about the services that you’re paying for and the value that the cloud provides.
Another thing is reducing IT infrastructure. When you move things to the public cloud, you should be reducing things on-prem. If you’re not doing that you’re paying more money in the aggregate. Several companies see significant benefits of moving from a CAPX model were purchasing all the hardware and software that they’re running on-prem can be burdensome for these companies and they end up moving to more of an OPX model where they’re renting those services. There can be significant gains through that, as well.
4. How do companies realize their cost savings?
Yeah, we’ve mentioned several things that can lead to higher costs in the cloud. If it’s a lift-and-shift situation, then you need to consider, “How can we right-size our solution so that we take advantage of the services available in the public cloud?” You need to be thinking about things like elastic scalability. You don’t want to put a solution up in the cloud that can deal with your highest load; you want to scale up and back from that highest load. What you need to think about is, “What’s the average solution that I need to put up in the cloud?” Also, once you’re there, you need to think about, “Are there parts of our existing solution that we would want to transition over to a consumption pricing model?”
You might rearchitect your system to take advantage of serverless and platform as a service solution in the public cloud. If you’re building things new, look at consumption pricing models as those are the ways that you can build things and save money right from day one. Serverless is the most obvious way to do that, so having a good understanding of what can be built serverless versus what needs to be built in a more traditional model. As we mentioned before, about avoiding hybrid solutions, if you have stuff running on-prem– if your complete software solution’s running on-prem and in the cloud, then that’s going to be the most sophisticated solution you can create and likely the most expensive.
And then, finally, a number of companies, when they say, “Hey, we’re going to move to the public cloud,” they open it up for everybody and all of a sudden they end up with all these workloads in the cloud and they’re like, “Why is this cloud so expensive? We don’t even have anything in production yet.” Well, that’s because you let everybody start running things and then they just left them running and now people don’t know what to do with all the stuff that’s running up there. Is it important? Can we kill it? Taking more of a policy-driven approach, understanding what needs you are, what areas you want to explore in the cloud, and then giving people the appropriate capabilities within the cloud and nothing more than that will help deal with that.
What you need to understand before you go to the cloud is what you want to do there, whom you want to allow to do those things, and making sure that you’ve set up a policy so that people can do what they need to do, but not anything more than that. So hopefully, this gives you some useful information to consider if you’re moving to the public cloud and how you can save money when you get there.