Challenges of Programmatically Reducing IT Costs – Dan Ortman, SoftwareOne

This is Textron tv. Hey guys, thanks for the throw. We're here with Dan Ortman, who is Global finops practice Director for Software One, and we're talking about finops and IT asset management and the current state of affairs with our overall IT economy.

Dan, welcome to the show. Hi, Mike. Thanks.

Glad to be here. I feel like somewhere along the way, spending on it just simply got outta control. I don't think we know where all our assets are.

I think there's a lot of things running in the cloud that people don't know how much they're paying for. And I guess my first question to you is, how did this standard of affairs come about? Yeah, great question.

Uh, you know, I think with with cloud, there's, there's sort of this balance, uh, we call it the iron triangle, but this balance between speed, performance, and cost. And the business is still asking it just like they have been for, for decades, um, to, to go fast. But the thing is, with cloud, we can go way faster.

You know, you know, engineers and architects that they have the ability to, you know, deploy, you know, instantly just about anything that they can imagine. So when, when speed as part of that equation, now all of a sudden you have access to, to deploy without getting a, a requisition for anything in in advance. So because you have that capability and the capability to deploy whatever size of a, a resource you need, or, uh, deploy even software without, um, going through procurement or I T A M, um, it, it, it's very easy for these costs to get outta control if there's no governance around that iron triangle and looking at, uh, what performance do we need?

When do we need it, and what are we willing to pay for it, what is the value we're gonna get out of it? And I, I think that that change is what's really caused the, the cost of, of cloud to, to sort of spiral. I think there's universal agreement that there's a need for change, but it seems like organizations are struggling with how to implement finops best practices and get their arms around it.

So what are the challenges that they encounter? Yeah, this is a good question. I I, I always like to, the, the very first part of the answer is the finops Foundation has done a survey, uh, for finops practitioners on what is the biggest challenge and three years in a row, uh, this, it's the same number one challenge.

And a lot of people have heard this line, but it's getting engineers to take action or it's changed to empowering engineers. So it's, it's not about blaming e engineers for, uh, you know, for the, the cloud spend because it's, you know, engineers are, are doing what the business is, is asking them. And, and so the, the, the very first thing in terms of best practices just getting started, uh, you know, the, the finops framework is, is three phases.

There's inform, optimize, and operate. And it's a, it's a continuous improvement cycle. And it, there's, you know, nobody starts with a perfect governance program.

Nobody starts with a perfect framework. So just getting started where you're, you're really understanding how to allocate costs, um, you know, where, where the spend is, you're sort of getting that informed phase down and then going and trying things, try savings plan strategies, try reservation strategies. You know, try, try these different tactics to, to get something started.

And then you can always improve. It's, it's, it's made to be, it's designed to be a continuous improvement framework. We see DevOps workflows everywhere, and I can't help but wonder if we cannot start to programmatically include some of the controls for finops into our workflows.

And so, you know, are we gonna see that from software engineers? Is that kind of the next step? Yes, uh, absolutely.

And I think one of the, the, the key tenets of the finops framework is cross collaboration. And a lot of the, the teams that I call out are finops needs to collaborate a across DevOps, finance, procurement, i a m security, you know, all of these different teams need to have a, a, a part to play in the, the finops framework. And DevOps is absolutely not an exception.

It's a, it's a team that should be involved in, um, helping to build, um, and influence what the finops team looks like inside an organization. And it absolutely should be part of the workflow as part of the process with a a, a DevOps team. And I, I would even take it a step further.

One of the things that the finops Foundation is, uh, has, uh, you know, built some, uh, working groups around is sustainability. And, uh, the DevOps teams that participate in the finops community also talk about sustainability, where it's even gamifying, not just who can build the, the most cost effective applications, but who's building the most sustainable workloads? Who's, who's using the most sustainable resources.

And so there's some really interesting ways that when you get, uh, the finops framework and um, the, the concepts around that CLO cross collaboration with within a DevOps team, there's some pretty cool things happening. Who's providing the adult supervision then? Who's kinda stepping up?

Is it coming out of the finance team? Is it this IT leadership? Who's the one waking up in the morning going, we got a problem?

Yeah, that's a great question. Finops is a weird thing in terms of where it starts. Uh, you know, we find that it's where who is feeling the pain first, it starts in all different parts of the, the organization.

But what is, but what is common is that you need somebody, uh, even if it's just a hat that they're wearing at the, at the time, you need somebody to sort of own that centralized function of, of finops. Sometimes that comes from somebody that might be in the cloud center of excellence. Sometimes it might come from somebody that's in IT finance, sometimes it comes from somebody that's in I T A M that has a passion for, for finops.

But at the end of the day, having that one person that really gets the framework and the, the, the concept around the cross collaboration, they drive that communication out to all of the different teams as well as driving that executive sponsorship. And, and that's one of the, the, the key things is without the executive sponsorship, then it's, it's really hard for any one of those teams to have success standing up the practice. Mm-hmm.

Of course, you can't walk down the street these days without somebody leaping out to tell you about their great new AI thing. So my question to you is, can we apply AI to solve some of these issues and save us from ourselves? Yeah.

This is, this is such an interesting topic. 'cause there's two sides to it. If you think of what it takes to, to build, not just, uh, you know, Microsoft co-pilot or you know, some of these things that might be subscription based, but if you're building apps on top of, uh, large language models and, and it's, it's your own data.

Just think of how much, how much storage, um, how much compute power that organizations are gonna be throwing at ai. It's gonna be very, some very cool things, don't get me wrong. But throwing all of that compute, all of that storage, uh, to, to make that go round, that's going to be expensive.

It's going to be a lot of resources, there's gonna be a lot of costs involved. So there's sort of two sides to the AI story. One is it's going to actually, in, in, in many cases, it's gonna significantly increase cost.

Uh, it's going to because of those things. But the flip side is you can also use AI in some cases to, um, drive some of those decisions. If you, if you have parameters around, um, you know, if, let's say for example, if you don't have, uh, you know, certain resources tagged or they're underutilized or things like that, and you can automate and you can use AI to discover and act on, on those things, AI can absolutely be one of those hats and on the finops team, so to speak.

So that, um, the, the things like getting engineers to take action and the, the manual intervention it takes in some cases to execute the savings that, that you can identify in a, in a strategy. If you can take that out and, and automate it and, and even have AI as part of that process, there's certainly value that can come from, from that as well. So it's, it's interesting 'cause there's, there's two sides to that coin for sure.

Where do I go and get the data that I can trust? 'cause I know cloud service providers will tell me how much they think things cost and, um, there's of course choices. I can still run things in an on-premise environment and sometimes that's more cost effective, but how do I go get some sort of, uh, baseline that I trust for the analysis?

Yeah. When it comes to finops and finops tooling, uh, we, we always talk about a, a tool belt. Um, there's, there's not really a, a silver bullet, but there are, um, a number of really good platforms that you can use.

Uh, some people choose to just use the, the cloud native tooling and data that you can get right from the hyperscalers. So you can, whether it's Microsoft, A W S G C P, um, you can get a, a lot of data directly from them and, and make decisions, uh, you know, that way. Um, there's third party tooling that have, uh, you know, amazing outta the box, uh, platforms where, um, you can not just have the data but have actionable recommendations based on the data that's coming out.

Um, and, and even going further into things like unit economics and, you know, what is the value that, that the business is getting out of, uh, those different workloads and, and, and things like that. So there's, there's a, a lot of tooling. Um, and there's some amazing tooling on the, the market that help to sort through what some of these invoices seem to be, um, you know, an insurmountable task, but literally millions, you know, we have customers where they're, their cloud invoices are millions of lines, and, and so having the right tooling, the right approach to that is, is is certainly critical.

Is it really about reducing costs? Because it seems to me a lot of the finance people I talk to, they just want some predictability. They want to know how much they're gonna spend from one month to the next, and it seems like it's all over the place.

So how do we just kinda maybe start with bringing some control into place so, you know, we know what to expect. At the very least, I would say everybody's in a different place. There are some organizations that are there.

Their mission right now is, is to cut costs. And so they'll ask us, you know, Hey, how can we cut costs by X percentage? Uh, are there's some organizations, like you said, or even certain roles that have different KPIs.

So somebody, like you said in finance, just having the, the visibility and, and, you know, accurate forecasting, uh, you know, predictability, that's, that may be a, a, a, you know, a goal for them. There might be others in the organization where, uh, you know, cutting cost is, is the most important thing. But we also find that as, as organizations mature through the finops practice, a lot of times it'll become less about cutting costs or even the predictability once, once they get better at that.

And it's more about back to that unit economic, um, discussion I was, I was having earlier where it's, what is the value we're getting out of this application, or what is the value we're getting outta this workload? And if we, um, you know, if we make these technical changes, let's say we lower the cost in a, by, by, you know, reducing certain workloads or having a better strategy around around the, the cost, um, does that increase the value of, of that workload? And if you can measure those unit economics, which might be, um, you know, might be subscribers, it might be, you know, in manufacturing it would be of course different than, um, you know, a, a an I S V or a a, you know, software developer.

So there's, there's different unit economics depending on what industry you're in, but being able to measure that against the business that has, uh, you know, a, a really huge impact on what is the priority, is it just predictability? Is it, what is the, the actual value we're getting from this cloud investment? Or is it, we know we have to spend the money, we just have to cut cost by x percent.

So help us figure that out. Is there a lot of wastage in the environments? It seems like, you know, you hear people talking about, you know, VMs that people forgot to turn off and, uh, they've been paying for and nobody's been using forever in a day.

So how much is there just an opportunity to save money by going after the wastage? Yeah, I've, I've seen a, a few different statistics on this, but the, so the interesting ones to me are even, um, not just from, from vendors that are saying, you know, 30 to 40% waste. There's actually studies done where organizations are saying they actually think that their waste is over 30%.

So I think the, the general industry standard, if you will, uh, you know, what's, what's accepted is what, where we're at today is, you know, there's over 30% of waste in the cloud. And we find that that can come in so many different ways. It, it might be storage and, you know, just not having an understanding of can we, can we change the storage tiering?

Can we get rid of it? Uh, you know, do we, do we need that? You, you know, that sort of thing to, uh, more advanced things like Kubernetes where, uh, you know, we may think, oh, we, you know, we deployed this workload in, in Kubernetes, it's a dynamic workload.

We, we've optimi optimized it, so you, you know, we're good to go. Well, the reality is, even in Kubernetes services, there's a, a, you know, a lot of workloads that are deployed in Kubernetes just with an unoptimized, kuber Kubernetes configuration. So, um, you know, we even find that there's significant waste even in a, you know, using dynamic technologies such as as Kubernetes.

So there, there's waste everywhere. And, um, you know, it's, it's very true. That's, that's, uh, one very consistent thing that we see.

And that waste, a lot of it seems to be driven to your point, by over provisioning. And we just have a bad habit of doing that because everybody thinks that there app is gonna crash, so they'll just provision it to the max. And do we need to have more faith in the automation controls we have?

Yeah, it's, it's true. And it's a, it's a great point. And, you know, I, I think, and this isn't to say that, uh, you know, again, back to engineers being irresponsible or bad corporate citizens, but you know, being a, you know, an engineer myself, if you asked me, you know, do you want the standard edition or the enterprise edition or, you know, do you want, um, you know, two cores or four cores?

You, you, you're always going to, um, error on the side of caution and, and overprovision rather than underprovision, because whatever you're building, you want it to perform well. You want it to, um, you know, be high quality, a good experience. And, and so you tend to, you know, if you given the option, you, you take the best.

And, and so that's, that's what fairly consistently happens without that governance, without the, the controls, uh, that you end up with an awful lot of over-provisioned resources. All right, folks, well, you heard it here. Over provisioning is still with us.

And guess what? You know, those cloud resources are not free, so use them responsibly. Hey, Dan, thanks for being on the show.

Thanks, Mike. All right. And back to you guys in the studio.