Cloud Cost Management: Embracing the FinOps Cycle

Jason Kelly:

Hello, I’m Jason Kelly, and I’m the director of FinOps at Anglepoint. I oversee our FinOps and cloud cost engineering organizations here. I’m excited to talk you through some of the basics of FinOps today. We’re going to cover topics such as what is the FinOps Cycle and challenges implementing FinOps, and how to overcome them.

So first, we’re going to talk about the FinOps Cycle. The FinOps cycle is something that you will see when you’re searching for information on FinOps, whether you’re looking on the FinOps Foundation website, checking out the FinOps book, or other resources. It’s a common graphic and a common description of how FinOps works, operating off of three phases: inform, optimize, and operate. These phases work in a cycle. They’re constantly going through this process and iterating and changing. You might be in the optimize phase in tagging, but maybe in the inform phase in budgeting and forecasting. This is similar to the FinOps maturity, which we talked about earlier, which was crawl, walk, run. This cycle can be different for every part of FinOps that you’re trying to do or every part of cloud engineering that you’re trying to do. So this doesn’t have to be one size fits all. You don’t have to be all in inform, all in optimize, all in operate.

So the informed phase is to drive accountability through data transparency. This is the beginning. This is where you start to get your reporting. You start to add accountability. You start to get basic things like tagging and other things like that. This is your phase where you’re starting to tell people what’s going on, giving them visibility into their cloud spend, their cloud usage, things like that. Optimize is where you start to take action on opportunities that you’ve found. So this is going to be low-hanging fruit. So, in terms of AWS, it’s going to be unattached DB volumes, it’s going to be right-sizing VMs and EC2 instances and compute instances. It’s going to be those kinds of things, but it’s going to move beyond that into rate discussion or fully modernizing a tech stack, things like that. Operate is the final phase of this cycle where you’re going to get into making heavy business decisions based on the outcomes from the first two. You’re going to start automating. You’re going to use AI. You’re going to use tools to help you manage your reserved instances or right-size your instances for your auto tag or things like that. So, this is the culmination. Once you have all the information, once you’ve done the optimization, you get into operate. And that’s where things start really, really flying. But again, it’s a cycle. So, once you get to the operate phase, you’re still going to want to go back to the informed phase. Are we now presenting our users with the right information based on all the decisions we made in the optimized and operate phase? Do they no longer need to know about these things that we’ve been reporting? They need new data. So you go through the cycle over and over again to update it and to keep it relevant to your business needs.

So, I’ll touch a bit more on the individual cycles themselves. So, the FinOps cycle in the inform section of the FinOps cycle is going to be things like your taxonomy. This is your tagging, your naming of resources, your account names, basically your accountability. It’s where you start determining who owns what and how you keep track of that. It’s going to be reporting and dashboards and simple alerting and things like that. It’s going to be the start of your show-back process. You can’t do show-back until you are charged back until you’ve got show-back. So this is the start of that. This gets you the foundation to get everyone looking at the data in roughly the same way, to get the people that need it in a timely manner and to make sure that your data is accurate.

The optimized phase, this is where you’re going to start looking at your ROI, your return on investment. So you’re going to start looking at the amount of resources you’re using and how you’re using them. You’re going to start looking at the pricing around that, whether it’s your contract negotiation with your cloud vendors and your tools, it’s going to be cost avoidance around reservations and spots and savings plans and things like that. It’s going to be decisions around which providers to use, which cloud provider to use for which tools. It’s going to be your ESG, your sustainability commitments and your responsibility there and where you place resources and how you use them. And then it’s going to be the return that you get on those, whether that’s environmental responsibility, whether that’s a return on financials whether it’s return on resources and the people dedicated to them. This phase focuses on all that. Are you optimizing all of this to get the best ROI that you can based on all these different factors, and this doesn’t have to be one calculation. This can be an ROI calculation based on very specific business units or very specific use cases and needs.

And finally, you get into your operate phase. This is where you’re using the decisions you’ve made and the reporting and the optimization you’ve done to start making strong, heavy business decisions on how to use your resources, where to put them. You know what technology to invest in what technology needs modernization. You start getting into things like really looking at decisions based on unit costs, more granular decisions that you couldn’t do until you had really good information, really good reporting out of your informed phase and really good optimization out of your optimized phase. And this is where you start getting into automation. This is where you start using things like AI and machine learning to make decisions for you. You start getting heavy-handed and shutting down resources that are non-compliant and wasteful. But you can’t get here until you’ve done the informed and optimized phase. You can’t jump to operate because you’re not going to be able to make the decision. You’re not going to have the information you need to make those decisions. And to really be smart in the business decision you’re making, the engineering decisions you’re making. So that’s the FinOps cycle, that’s inform, optimize, operate again as a cycle. And it totally depends on where you are in your engineering journey, your cloud journey as to where you’re going to fit within that cycle.

So now we’re going to talk about challenges and solutions to implementing FinOps or to getting a mature FinOps organization. So first we’ll talk about some of the challenges in implementing FinOps. One of the big challenges is just cloud cost management. It’s different. It is complex, it is complicated. There’s a lot going on and things changing every single day. It is not like a physical data center in which you own the asset. It depreciates and that’s pretty much it. Every service, every resource you use has a different cost associated with it. And if you’re using a pay-as-you-go model or an on-demand pricing model, it seems like those prices are never the same. This can appear to be a challenge. It makes it difficult to get accurate forecasting and resource allocation like you would get in a data center.

It appears that you don’t know what’s going on in cloud and that it’s difficult to predict. It’s difficult to forecast. But a lot of that comes from just a lack of understanding of really how to accurately forecast, accurately budget and predict what’s going on in the cloud.

Next, you’ve got organizational conflicts. So you’ve got departments like your cloud engineering, your DevOps, your cloud infrastructure teams all working in one silo. And then you’ve maybe got traditional data center engineers or traditional engineers or service engineers or procurement and others working in a different silo, and they often don’t want to work together. Engineering, cloud engineering is often worried about losing its agility, losing its ability to do what it needs to do in the cloud when it needs to do it. So you get these conflicts between these departments that don’t want to talk to each other because they’re afraid of losing their autonomy. You’ve got resistance. Just, in general, companies are usually resistant to cultural and process changes. They don’t want to do something new. It’s difficult. It’s something new to learn whatever it is. So you get these conflicts that arise from that. And then you also get just challenges with getting executive buy-in. FinOps is not free. But its return on investment is huge, but getting an upfront cost paid for requires executive buy-in so convincing executives that it’s worth the time when there is a lot of fear and uncertainty and doubt when it comes to cloud, when it comes to FinOps and how to operate resources there, it can be challenging to get an executive buy-in. And then finally, workflow integration. FinOps requires a huge amount of alignment with. Other teams with procurement, with asset management, with security, with cloud engineering, there’s a lot of integration that has to happen. And it’s not just integrating with your Jenkins or integrating with your deployment of Terraform or your cloud formation or whatever it is that you’re using to manage all this. It’s integrating within the process, getting in on procurement meetings that impact cloud, getting in on contract negotiations, things like that. These can all be a challenge in implementing FinOps.

So next we’ll talk about some of the strategies for successfully deploying and how we can overcome some of that. One of the first strategies is iterative adoption and quick wins. So have a phased approach to your FinOps implementation. Don’t try to boil the ocean. Go take off small chunks. Find small teams that are willing to implement FinOps best practices and use them as champions. Use them as wins for the company and showcase them, get them to espouse the wonders of FinOps and how it got them to better accountability, better predictability around what they were doing in the cloud. And also identify early, major cost savings opportunities. There’s always low hanging fruit, and those can add to very quick wins and a very quick realization of the upfront cost of implementing FinOps. If you can go in and remove low hanging fruits like unattached volumes or databases that haven’t had connections with 30 days or something like that, you can save a significant amount of money right away with very little work having to be done, to realize those and those, can, help you set up your FinOps for success long term. And then you use those as a springboard to then look at larger cost savings opportunities to show. That it wasn’t as much engineering effort. It wasn’t as much time as originally thought. Or maybe there’s this perception of in trying to save money.

Your next strategy is going to be implementing a cloud center of excellence. So this is going to be a team, a centralized team of experts that come from finance, cloud engineering, security ops, a business office, it’s going to be a group of people that are well connected to all these different spokes within your business and can help share not just best practices and expertise when implementing new services or making major decisions, but are connected and know what’s happening so that when you say. When your license management team decides that they don’t want SQL Server licenses anymore through Microsoft, that they’re telling the FinOps team that’s going to happen in six months. So the FinOps team doesn’t go out and buy reserved instances for the EC2 instances that are running SQL Server and now beholden to a three year contract on something they’re not using anymore. Their CCOE helps facilitate those conversations. It also creates continuous education and external expert engagement around what’s happening in the cloud. You can train your finance people on FinOps directly or indirectly just having in the same room and talk about the challenges that the FinOps team is facing in the CCOE meetings. Your finance people will now understand a little bit more about what’s going on and why what they do is important to cloud engineering and FinOps and whatnot, and that works amongst all the teams.

Your next is going to be your cloud management platform. These are going to be your tools, your reporting, your automation, things like that. It’s your exposing of where your cloud costs are. And most cloud providers provide these out of the box, whether it’s your cost explorer in AWS, or it’s your budgeting tools or your billing reports, whatever it is, there are tools out of the box that you can very quickly implement to show what your cost is, where your resources are being spent, who’s spending them, all these kinds of things that very quickly give you insight into your cloud costs, your cloud usage and use those and then mature into maybe third-party tools or building your own tools to help surface those costs, to bring in the accountability metrics, whether it’s tagging or organizational charts or some third-party Metadata that you have or your own metadata that you have to help create accountability around cloud resources.

Use these cloud management platforms to bring all of that in and answer the questions that are being asked, not the questions you think should be answered. And then finally is engineering engagement. So we talked about this with iterative and phased approach to FinOps. This is very similar. Engage with engineers. Encourage cost ownership among them. Find the people on teams that really understand that everything they do in the cloud has a cost associated with it and hoist them. Get them to be your champions on the team. Enable them to answer the questions that they have and show the rest of their teams and other teams how they’ve been able to drive their costs down with very little work on their part and then get them bought into this idea of shared financial accountability. Even if they’re not doing it purposely, even subconsciously they’re starting to understand they’re accountable for what’s being done in the cloud. They make decisions around new services that they deploy and how they build new applications and new services in the cloud based on. The cost associated with the cloud, the efficiency in the cloud.

So those are our challenges and solutions for those challenges in implementing FinOps. If you have any questions, reach out to us.