John Platt, VP, Machine Learning at StormForge, was a guest on TFiR Insights “Let’s Talk”, to discuss the 2021 Cloud Waste Survey Report that points at a growing global problem of cloud waste.

In this episode, host Swapnil Bhartiya and John discussed a wide range of topics including the factors that lead to cloud waste and the real impact it has on businesses, security and efficiency of operations. Of course cost is a big factor, but there are many other factors that organizations should be aware of as well. They also talked about the ways organizations can contain their cloud waste and use resources more efficiently.

For more information about Cloud Waste, read our Cloud Waste Survey Report.

Read the Report

Swapnil Bhartiya: Welcome to TFiR Insights and I’m your host Swapnil Bhartiya. StormForge recently published its 2021 cloud-based survey report and to discuss this report we have with us John Platt, VP of Machine Learning at StormForge. John, it’s great to have you on the show.

John Platt: Thanks for having me.

Swapnil Bhartiya: Before we talk about the report, I would love to learn a bit about the company itself. Tell us what are you guys doing there?

John Platt: So, StormForge is all about optimizing the cloud resources so that you can always make sure you have the right resources at the right time for each of your applications, meaning that you always have good performance, but at a reasonable cost. 

Swapnil Bhartiya: If I’m not wrong, your focus is more on reducing cloud waste? 

John Platt: Exactly, exactly. And that’s where the cost angle comes into it. With the move to the public cloud, cloud waste is out of control and it has a major impact on enterprises. Lower profitability, reduced competitiveness, and also carries reputational risk for IT organizations inside sort of large enterprises making them seem wasteful, inefficient.

Swapnil Bhartiya: In general, it may seem like hey because it’s so easy to just put your credit card in there and you know get resources quickly so that you don’t even know you know that you’re paying for it. Is that the factor, or there are other factors that contribute to cloud waste?

John Platt: That’s definitely a factor. Especially as sort of purchasing compute and cloud resources moves from centralized financing departments and now individual developers can spin up resources at a single click, but the two big drivers of cloud waste – it’s typically complexity and over provisioning. The cloud is an extremely dynamic, complex environment, which makes it very hard to get right. When working with applications often you don’t know exactly what resources they need. So you end up being caught between providing too few, in which case you’re going to hurt customer experience, or you’re going to provision too many and then you’ll blow budgets, but the customer will be happy. When caught between these two options, everyone picks the customer, which is how you end up with widespread overprovisioning. With kubernetes this gets worse because it’s more dynamic, it’s more complex, you’re provisioning resources at a much more granular level and you’re sharing resources between lots of containers running on a node. So it’s the usual problem of over-provisioning and complexity in the cloud, just 10x harder.

Swapnil Bhartiya: You did allude to that, you touched upon a lot of those things know that what does it mean you know of this cloud waste because of course you are not that efficient with your resources, particularly financial resources, but what are the other even more significant impact of cloud waste for the enterprise customers? Also let’s just quickly think about those cloud vendors also. Of course they’re hyper cloud providers, so they don’t care about how many dedicated resources. It can be a case in the VPS environment where every node matters, but from enterprises perspective what is the impact of cloud waste?

John Platt: So the impact is seen on the bottom line. 50% of cloud spend is wasted, which means people are spending twice what they need to spend. So this impacts profitability, it impacts competitiveness, and it impacts the business as a whole because you’re wasting money on cloud resources, which you could be using to hire more engineers, more people, shipping more features, more products etc.

Swapnil Bhartiya: If you look at it in the kind of dollar amount, how much percent of resources are companies wasting just on this cloud waste that money they could have used in R&D and so many other areas?

John Platt: So typically these cloud bills can be extremely high – tens of millions per month. You can find the exact details and the breakdown in the survey which you mentioned at the beginning. So 50% of that. If you’re spending millions per month then your and half of it is wasted, you’re wasting millions per month.

Swapnil Bhartiya: I also want to understand, sometimes we focus too much on technology, but in most cases it also becomes people problem, cultural problem. So when we do look at cloud waste, just like e-waste, is there any cultural aspect, people aspect to it as well?

John Platt: 100%. So this is a big part of FinOps, a new foundation which is trying an unused movement, which is trying to focus on spending and spending money effectively in the cloud. It’s very much a cultural problem and it’s bringing together people who wouldn’t typically talk. Engineers, finance, product owners, everyone. So it’s very much a cultural thing and about breaking down barriers and it’s not just a technical discussion. It’s also sort of how do you share information, how do you understand who’s spending what, and how does that tie to business objectives and is the return on the investment there.

Swapnil Bhartiya: This may not make much sense in this case, but if you look at the traditional data centers or, you know, if your light is running, your light is running. If you have 100 cores of CPU, one gigabyte of RAM, it is there you know whether you use it or not, power is on, power is on. That is not the case in the cloud, so I just want to go back to the cultural aspect as well, but because when we do look at cloud, we are not looking at the legacy workforce who do not understand, or is that problem still there that people are not even aware that you are paying by use, you’re not paying for the whole rack of servers there.

John Platt: So the the on-demand nature of the cloud definitely makes it easy for things to get out of hand quickly because when you turn it on you’re paying for it and if you forget to turn it off you’re still paying for it. I think it comes down to sort of two ways of actually reducing resources. It’s first realizing that if you use two CPUs less tomorrow, then it actually impacts the bottom line immediately because of the on-demand nature of the cloud, and it’s about sort of being very careful about assessing where I can actually go in and it typically falls into two camps. There’s either things that people have started and forgotten about, and there’s just orphaned workloads. Then the second one is there’s important workloads, which are just over provisioned. People don’t know exactly how many resources they need, so they gave them too many. 

Swapnil Bhartiya: We talked about the problem aspect of cloud waste. Let’s look at the solution side of it. Number one, I want to understand from you is that do you see any trend, any patterns where people are getting aware of it and they are making even if you know small steps towards reducing a cloud waste, or you feel that a lot of awareness is needed?

John Platt: So there’s definitely people making progress on this front. One thing we think is a really natural solution here is machine learning. So one of the reasons, even when people know that they’re over provisioned, one of the reasons that they don’t fix the problem is because it takes too much time to go around and have engineers manually right-size applications. With machine learning, you can watch a lot of the monitoring data coming out of these cloud platforms and you can actually automatically right-size. So people get all of the benefits without having to waste a lot of engineering time and then a lot of small wins add up to a lot of big savings.

Swapnil Bhartiya: Earlier you mentioned FinOps Foundation. Can you also talk about what is the role that FinOps Foundation is playing this space and how involved are are you or StormForge is with that?

John Platt: Yeah, so the FinOps Foundation is a new movement trying to encourage people to start thinking about how they’re spending money in the cloud. Rather than it being just people spin them up and then never go back and check was it a worthwhile investment, actually encouraging people to regularly understand what they’re spending, see if that makes sense, see where they can save money, and just making it part of continuously sort of optimizing costs, rather than just waiting and getting a huge bill at the end of the month, the quarter, the year.

Swapnil Bhartiya: Before we wrap this up, I think very, very timely question is that because of this pandemic, a lot of users are rushing towards cloud, digital transformation has been accelerated, which I’m pretty sure will also lead to more cloud waste. Do you see any spike in cloud waste in coming time or do you think that these users are already well aware a lot of education is going on there? StormForge is coming out with support that will help them. What kind of trends do you see in future?

John Platt: So the trends I see are the following. So first of all the drivers of cloud waste aren’t going away. These cloud platforms will always be complex. They’ll always be hard to run on and the on-demand nature will always mean it’s easy to spend a lot of money very quickly. With the move towards the cloud, public cloud is only growing. It means cloud waste will also only grow. 75% of people realize that this is a major problem in their enterprise, and everyone expects their cloud spend to grow. So with the core drivers staying the same and the sort of industry trends moving towards the public cloud, this is a big problem which will only get bigger.

Swapnil Bhartiya: Awesome, John. Thank you so much for taking time out today to talk about this report and also kind of deep dive into the very serious problem of cloud waste and the solutions that are there. I love this conversation and I would love to have you back on the show. Once again, thank you. 

John Platt: Thanks for having me I’d love to come back.