The past two years have seen erratic cloud spend thanks to the upheaval caused by the global pandemic.
Looking ahead to 2022, business is beginning to normalize. Within that “new normal,” there is no question that cloud is more popular than ever, and Kubernetes is at an all-time high. Some companies are continuing to operate remote, or find that they are thriving thanks to an investment in the cloud.
At CloudZero, we believe that every cloud investment should be a smart and efficient investment.
Often, cloud costs are a source of misunderstanding between finance and engineering. Finance wants to understand what is driving cost, while engineering wants to build innovative solutions. The problem is, they haven’t quite figured out a common language to communicate.
But developing a common language results in common buy-in and allows for building more efficient and cost-effective cloud systems.
Kevin Taylor is the Head of Cloud Services for Exstratus, a leader in the cloud cost-optimization space since 2015. He sat down with CloudZero Founder and CTO Erik Peterson to layout current trends and their top five cloud predictions for 2022.
Current Cloud Trends
When it comes to emerging cloud trends, Kevin and Erik identified four key buckets.
- In DevOps, if you build it, you run it
- The rise of improved cloud economics and visibility
- The rise of cloud-native solutions
- Increased vendors for cloud cost management
Since the dawn of the cloud, there has been an ongoing conversation about the role played by DevOps. Cloud services gave rise to a new DevOps mantra of “you build it, you run it.” Ultimately, increased accountability in the DevOps realm is a positive trend that enables smart solutions.
With cloud usage increasing for most companies each year, the logical next question is: are we getting a return on our cloud investment? Running a good business means letting the health of your business influence your cloud decisions.
Understanding your ROI becomes easier when your architecture is set up effectively (e.g., you are cloud-native). For many companies, this means thinking of your cloud setup as an operating system that leverages native API, removing dependencies so you can more easily scale up or down.
“We all know if you build native right native code for the platform that you’re going after, you’re going to get a better return on that. You’re going to have a better experience,” says Erik. “When I think of cloud-native, it’s, am I leveraging this amazing innovation like it’s an operating system? And am I building natively for that operating system?”
Finally, vendors are reading the writing on the wall and recognize that these areas are ripe for growth.
“The amount of vendors who are playing in the cloud economics space, including some who came in to have a niche with things like Kubernetes cost management, the growth was crazy,” says Kevin. “People don’t just get into the business unless they think there’s business there to be captured.”
Analyzing the current environment as a whole resulted in Kevin and Erik making the following five predictions.
Prediction #1: Engineers Will Be Held More Responsible
There is no truer statement as it relates to cloud engineering today than this: every engineering decision is a buying decision.
“Your software engineering teams, they’re the ones actually making decisions about how your bill is going to pan out at the end of the month. Sure, you could be buying RI’s and savings plans and trying to get out ahead of that, but that only gets you so far. That’s what the first generation of cloud cost management was all about, was let’s try to mitigate what the teams have built. Now, it’s about what you’ve built,” Erik says.
The key is to take steps to make sure that engineers understand the cost of their actions. This happens by feeding them essential data.
“I don’t know any software engineer who, given the right amount of data, won’t think and do the right thing. And this doesn’t slow them down. What we’ve seen is that this empowers them to actually build more efficient systems and innovate faster, because they now feel empowered. They have the data at their fingertips,” says Erik.
Ultimately, engineers are making cloud buying decisions as a result of their daily actions. It is, therefore, no surprise that engineers will likely continue to be held more accountable.
“I think that is the number one reason this is going to become a hotter trend is because ultimately some of the things that caused this dysfunction between the orgs is that the finance people don’t know how to ground this stuff in the daily workflow of engineers,” says Kevin.
This is important, because if decisions are made in an ad-hoc and piecemeal way, cost allocation and cost attribution become more challenging over time. Cost allocation is hard because you need one thing: consistency.
Innovations in the review process for actions like pull requests are making it easier to add comments, and therefore accountability.
“When you’ve got a review process, this is the number one way engineering teams kind of enforce standards and hold each other accountable,” Kevin says.
Prediction #2: Cloud Cost Will Become A Board-Level Issue
It’s common to see public companies getting pulled through the wringer when they do their filing. Everybody scrutinizes their cloud commit and talks about how much money they’re spending.
“We have to have that same level of scrutiny from the board down for what it costs to deliver value in a cloud-driven era. And we have to know, and the board wants to know, is that being done efficiently? That’s why I think it’s a board-level issue,” says Erik.
The challenge will become staring down that scrutiny and having clear answers for why cloud costs are going up.
Kevin says that key factor here is the cost of goods sold.
Imagine you are a SaaS business and you sell more than one product or package. Ultimately, what the board wants to know is the margin on each.
“This is why things like the ability to do cost allocation matters. Because if you can’t attribute what costs go to what product, you can never answer the question for the board: what is our margin? And it makes it also really hard to do forecasts,” says Kevin.
While you can get away with playing fast and loose in the early days of a company, that grace period quickly ends. As soon as you go to seek funding or undergo SOC readiness, those questions are going to be asked.
That’s why cross-departmental conversations between engineering, finance, and executives are so important. Shared services need to be truly shared so that they are visible and effective to all parties.
Prediction #3: Unit Economics Will Make A Bigger Impact
It’s not enough to understand what you’re spending on cloud services. You need to understand what it costs to deliver value. To do so, you need to have a firm grasp of your unit economics to service a customer.
“I think the companies that are starting to think very strategically about this spend and realizing it’s not about cost, it’s actually about managing your investment. Because if I’m successful, my cloud cost is actually going to go up. If you’re a successful SaaS, your cloud costs are going to go up. And I wish that on everyone. What I don’t wish on you is that your unit costs go up at the same time,” says Erik.
You should be seeing that trend remains flat or even decreases over time.
“If your business is successful in growing, the cloud bill is going to go up. This is predictable. But if you don’t have some way of boiling down whether or not these growing costs are correlated with growth or they’re an erosion of the bottom line, it makes it really hard,” says Kevin.
Your goal should be to avoid “panic months:” months when executives spot a huge jump in cloud spend, and there is no clear explanation of whether costs increased because revenue jumped, or if cloud costs increased alone.
“I think this is why your unit economics matters. Because it gives one of the best level playing fields to talk about things in terms of letting executives understand is this a problem or is this not at all a problem?,” Kevin says.
Unit cost metrics are a common language that engineers can use to make better decisions. For example, perhaps a specific API call can be made 10 times per day versus hundreds of times per day, lowering costs and turning an engineer into a hero with one line of edited code.
Prediction #4: We Will See A Shift In Cloud Cost Optimization
The future of the cloud isn’t going to be about buying better solutions, but building them.
At first glance, it can be hard to understand why companies are using things like Kubernetes. Kevin says that it comes down to smart architecture.
“When you start thinking of it in terms of what architectural differences are there, things like ‘Oh, I have workloads that I’m running BM’s for 24 hours a day, but really they only need to run when an event comes in.’ And I scale that to zero, because I have a cluster that’s always there and the control plane is always there. And then only spit it back up when let’s say a new ETL job comes in, or a new request comes in for a batch. I think these are the sort of things that are going to become important moving forward,” Kevin says.
You also want to architect a solution that is imminently flexible, allowing you to scale up or down. Most companies only consider scaling up, but events like the recent pandemic are a cause to pause and consider the opposite need.
“A lot of people hadn’t thought about scaling down. And that really gets at that building better idea. If you’re ultimately building a really elastic system, it goes up, comes down. Responds to events, responds to load, responds to demand, getting closer to that curve of usage. And that’s exactly why a lot of people got on the serverless and the Kubernetes or containerization bandwagon alike. Because they wanted to map to that consumption curve,” Erik says.
Ultimately, you want to position yourself to both make better buying decisions and build better.
Here are the 3 most common questions we get about cloud cost intelligence.
Prediction #5: Say Goodbye To Legacy Cloud Management
More complex cloud solutions and platforms require more sophisticated management. The days when you pull your billing reports and agonize over them side-by-side with your cloud metrics will likely soon be over.
“Businesses don’t stay static. If you’re in a hybrid cloud environment, eventually you need to have a unified conversation. These sort of things kind of leave a lot of that conversation up to a person trying to pull this data together on a spreadsheet. If you are going to multi-cloud, you also need something that means that you don’t have to do all of this work times number of clouds and then have somebody overlay it and put it onto a single dashboard,” Kevin says.
New cloud management will help manage these complexities in the face of business needs. What’s called for are dynamic, real-time solutions.
“I see cost data as telemetry. It’s a real-time event about some activity in the environment. And combining that telemetry with events as it relates to how your systems are operating. How do I correlate those things together? How do I pull that stuff together is really important to get at that unit cost conversation.” says Erik.
For example, if you want to know your cost per 1,000 emails processed, you need to know how many emails are being processed per customer and correlate that with how much money you are spending to deliver the email processing part of your system.
It would be much more effective to use a platform that combines those factors automatically, versus exporting data from multiple systems, gluing it together in Excel, and then trying to generate some nice reports — all when an exec needs that data right now.
“Shared costs should not be built on static rules. It should be using real-time telemetry so that you have a real understanding of what the impact is. Traditional tools just can’t do that. They were built as reporting tools, not event-driven, data-driven platforms,” Erik says.
At CloudZero, we use our own platform to gain a deep understanding of important metrics such as our cost per customer. When we onboard a new prospect, we think about questions like:
- What does it cost to operate this?
- How do we price it?
- How do we manage our systems?
- Where do we apply our engineering dollars?
Our customers see the same thing and benefit from leveraging the same information.
to see personalized recommendations and gain insight into how CloudZero can help take your cloud operation to the next level.
The Modern Guide To Managing Cloud Costs
Traditional cost management is broken. Here's how to fix it.
