By Nishant Gautam, Head of Cloud Financial Management, AWS India
A 2020 study from 451 research with US and UK customers, looked at their cloud cost management practices, 80% of the respondents acknowledged that they were doing a poor job of managing the cloud, a poor job at managing cloud financial management and that this had a negative impact on their business.
It was slowing their cloud adoption, crippling innovation, increased costs among others. Flexera state of cloud report 2022 suggests that Cost Optimization and financial management in the cloud is one of the most important priorities. (it has been ranked as the top priority for the 6th year in a row)
Hence, there is a need for a FinOps strategy in place.
There are many reasons why the cloud is different from traditional data centers:
1/ Cloud offers on-demand provisioning that makes it easier for anybody to spin up resources. So, there is a need to think about the guard rails that are needed to ensure that only those individuals or teams that need to do that job are able to do it.
2/ Public cloud providers offer many services (for example, AWS offers more than 200 services). Each of those services has its own resource types, different pricing model, and whatnot. So, while the pricing model is transparent, navigating the landscape of the different pricing models can be complex even for advanced users. So, this is something that needs appropriate attention.
3/ Our studies indicate that customers who moved to the cloud oftentimes bring over-provisioned on-premises infrastructure sizes with them. They need to break the habit of over-provisioning their infrastructure, especially production environments. Customers will benefit most from the cloud when you pick what is required, and how much is required. “Biggest is not the best” in the cloud. Also, remember that resizing in the cloud can be challenging. Sometimes this means you have to take an outage, we have to turn that service or that server off for a few seconds or a minute to restart it.
4/ Forecasting costs are more complex, given the elastic usage patterns associated with the dynamic workloads and pricing models.
5/ when resources are provisioned, often you’ll see this with migrations, owners of an application may not have any visibility into what their applications are actually costing. They may not have any visibility of what they’re costing them on-premises and they may not have visibility into what they cost on the cloud. So, technology organizations may not be historically attuned to looking at the cost of things as a first-order non-functional requirement. So, as an engineer, one is used to looking at performance requirements and security requirements, and scalability requirements, not necessarily looking at cost requirements because he/she is probably using infrastructure that had been purchased already in the data center. And so, getting engineering teams to really start to look at that, and get an awareness of resource costs in mind is a critical thing as well.
6/ And finally, even after cloud waste is identified and the dynamic nature of cloud consumption is addressed – the fact that the cloud services are constantly being revised and introduced means that we need to do cost governance in a continuous manner. This isn’t a static thing.
So, there are lots of ways in which the cloud is different than traditional data centers and we have to be ready to change those processes. A mental model that may help with the transition is that the cloud is another data center, but one that requires organization to evolve their existing financial governance capabilities.
Having FinOps / Cloud Financial Management capabilities means maximum value realization. The purpose is to evolve the customer’s cloud consumption model and think about the capabilities that the customer needs to develop to manage that cloud cost-effectively over time. We’re going to talk about FinOps capabilities in terms of the PLAN, RUN, SEE and SAVE pillars.
PLAN – The plan pillars are the planning and forecasting arm and that’s where we need to look at how we currently plan to use the cloud in the future. How are we spending? what are we spending on? I may know that next quarter I’m going to have a giant testing phase for a new module or a new capability that’s being launched. And so, I may have to budget for the cost in that way. Whereas in a traditional data center model, I would have probably purchased all that testing hardware up-front, but in the cloud, we have to plan and budget for that effect.
RUN phase is really your cloud financial operations – And that’s where we’re looking to have the right people – roles/responsibilities, the right process, and the right technologies – tools/automation in place. This is less about the cloud and more about your business. This is about thoughts of how things will work day to day operationally once we start using the cloud.
SEE – So SEE is essentially our measurement and accountability. We want to see what we’re using. We want to be able to report on what I’m using. Sub-capabilities of the SEE pillar are Tagging Strategy, Accounting strategy, Show back / charge back capability, and right KPIs.
SAVE – In the save pillar, I want to see what cost optimization levers, I’m currently using to optimize my spending. We will analyze and look for opportunities to improve the amount of spending on the cloud so that we’re only spending what we need when we need it.
These activities and capabilities are actually very similar to what you already do for on-premises environments. But instead of performing them in a macro manner, we’ll need to apply them in a more micro way, continuously for many of the reasons we mentioned earlier.
To summarize, one should have a FinOps / CFM framework that organizations can refer to not only to create visibility and optimize cloud spending, but to make cloud spending more predictable for the business and stakeholders, and build a long-term, self-sustaining cost-aware culture supporting all waypoints in the cloud journey.