#054 – From Shiny Objects to FinOps: Taming Cloud Costs in the AI Era with Josh Schlanger (CloudXray AI)

Itiel Shwartz: Hello everyone and welcome to another episode of the Kubernetes for Humans podcast. Today with me in the show we have Josh. Josh, happy to meet you.

Josh Schlanger: Nice to meet you as well. Glad to be here.

Itiel Shwartz: I’m happy that you are here as well. Maybe you can tell us a bit about yourself, like what do you do?

Josh Schlanger: Yeah, I don’t know if we have enough time in this whole podcast, but I’ll try to keep it short. So, yeah, so I’m Josh Schlanger. I’m based here in the US out of the Bay Area and I’ve been in I say I joke I’ve been in industry a long time. I used to have hair if you’re watching this. I used to have hair when I started. don’t so much anymore. But the last I would say 15 years or so in my career I’ve been primarily focused uh in infrastructure, DevOps, leading DevOps teams, platform engineering teams, SRE, owning FinOps and all the things associated with cost in that space. worked in different verticals where I worked for MarTech companies, worked for e-commerce companies, worked for health tech companies. So, had a pretty large span of companies, you know, different types of companies I worked for. All, you know, trying to solve their business problems with the, you know, the similar technology. So, I’ve been able to bring my knowledge and experience to those companies. And then, you know, recently I uh left uh the company I was working at. I decided to want to go out on my own. So, I’m hyper-focused on FinOps and particularly to doing two things, you know, consulting and advising with companies to help them get their costs under control as well as as part of that process as I learned that I don’t know from a FinOps perspective there’s lots and lots of tools and and service companies in the space. There’s even I started building a uh a database a directory I call it of FinOps companies. I’m tracking close to over 200 uh so far that I’m talking to. So, anyway, that’s a little bit about me uh and and what I do and some of my experience.

Itiel Shwartz: No, it’s like a great intro. So, maybe you know, you have this you have all of this background different sector which is cool because we meet a lot of the time people who are in the same sector. Maybe maybe let’s start with that. Just, you know, if you can talk a bit about the difference that you had as someone who is in charge of DevOps SRE in different sectors. I think it can be quite interesting to our listeners.

Josh Schlanger: Yeah, sure. You know, it it’s interesting, you know, for me, you know, the technology was always the same in those different companies, right? Now, the technology changed, right? It’s dynamic. They’re adding new capabilities. I mean, I remember the day when Kubernetes was first released, you know, from Google and then people started using it and we I was at a MarTech company and we considered using it, but it was way too complicated. There weren’t a lot of tools. You had

Itiel Shwartz: MarTech is like marketing tech? Just so I

Josh Schlanger: Yeah, MarTech Yeah, sorry. Yes. So, marketing technology companies.

Itiel Shwartz: sorry. We No, yeah, so for the audience of MarTech marketing technology companies, SaaS-based solution. and you know, in that company we looked at Kubernetes in the very very early days. We said, “No, you know, we’re a small team, too complicated. Let’s look at something else.” So, we we definitely went container route. We used a tool called Rancher. That was great for us at the time, but the point being is it wasn’t about the technologies in those different companies because you can you can go from different verticals to different verticals. It was about what’s the business problem you’re trying to solve and then will the technology fit or solve that business problem. And in a lot of cases from a scale perspective, from a growth perspective, you know, those technologies we adopted were useful. We Yeah, you do have to be cognizant of not getting caught up and I’ve earlier in my career I got caught up in this problem is like you love technology like, “Ooh, here’s a new bell with a shiny object, right?” And you have to be very careful. It’s like, “Okay, let’s really solve the problem. If we can use the new shiny bell, that’s great. If not, let’s use something tried and true.” Okay, so that was like the MarTech. What about like uh e-commerce and like uh like other other areas? Like, you know, like we don’t need to go over all of them, but which like area is the best for like DevOps people or like operation people if any?

Josh Schlanger: I I think they’re all great. I don’t think it really matter, you know, I mean, it it’s a bit of a you could call it a bit of like, you know, ride the ride the middle here, but really any of them have technical problems they need to solve and they need DevOps platform engineers SREs to go solve them. You know, I think it becomes a little bit of a personal preference, you know, what are you interested in as part of that solving that business problem where you can apply the tech. Are you interested in e-commerce? Do you feel think it’s interesting? Now, I worked for um uh the company did online consignment. So, people would consign their items. I actually found that whole workflow super interesting where people would send their stuff into a warehouse. I went to the warehouse. I saw the whole process and how we use technology to get everything in, to then posted to the site, through the sales process and then shipping out. Right? And there was all kinds of unique things that we were doing for that. But again, at the same same time we were the technologies behind the scenes and the infrastructure we were using was the same as what I’d used previously. But I found that one particularly interesting. The health tech one, you know, again, another super interesting, you know, challenge. How do you connect patients to um the right providers? in this case it was mental health focused. Right? And solving that problem. We had different challenges, but we were using again the same technology. So, I think it becomes, you know, it’s a little of a cop out. I think it becomes a personal decision on the types of companies you go work for, but the technology at the core is still the same. and and you still need to apply it in in a way that makes sense for that business.

Itiel Shwartz: Okay, it’s like a bit like politician answer, but but I let it slide that that you like all this really same.

Josh Schlanger: I have more opinions on uh going forward. How about that?

Itiel Shwartz: Yeah, no, no, no. Like that’s that’s your your like um your call. So, maybe like I ask you a different question. Like, what industry spend the most money without like any real need? Cuz now what you do is like cost optimization, right? So, are there different areas? Maybe maybe like share a bit about your journey to like this FinOps space even before that. And then I’ll ask a couple of follow-up questions on on that topic.

Josh Schlanger: Yeah, sure. So, you know, I joked in the beginning I you know, I started industry with hair. and I started industry before cloud existed, right? So, it was all data centers, you know, closets in in offices where you had servers, you’d have to buy software. So, it was driven by capital capital expense or CapEx budgets. And a lot of times you’d have to go out to the CFO uh or the CTO and CFO and say, “Hey, look, we need we need to go make this investment. We’re doing all this. We’re growing. You got some product roadmap. Here’s what we need to go buy.” Okay, and then we’d figure out how best to use that. And I enjoyed that aspect of the business. And then when things moved to cloud, there was it wasn’t CapEx because you push your credit card down, right? And it became all operating expense or OpEx. And now though, it was a little bit more wild wild west because people would spin stuff up, they would forget about it. You know, it would So, it caused all kinds of challenges. And getting wrapping your hands around that and getting control of that was really super super critical and important, particularly for SaaS companies because a lot of that expense impacted your margin. So, I may talk a little bit Yeah, so for those who who may not know who are listening, right? You know, we’re talking about really the gross margin, right? So, your top line revenue less your cost of goods sold or cost of sales, whatever that number is, it becomes your margin number. So, making sure that you’re in a SaaS business you had to hit a certain percentage, typically in the 70-80% range. Well, if you’re running resources in production that you don’t really need, you could potentially impact those numbers. And I’ve seen it happen. I’ve seen it uh go down to, you know, 15-60% which would be it was a real problem. So, how do you bring that back up? So, again, I always enjoyed that because it was a it was a it was a problem. I love puzzles, right? So, how do we solve this problem? How do we deliver what we need to deliver for the customers, making sure performance and reliability was there while also managing the costs. And uh so for the companies, you know, that I worked in in these spaces that I talked about previously, e-commerce, MarTech, it was always like, “Okay, we’re going to go build this, but let’s do this in a in a smart and intelligent way.” And then use the tools in the market to either give us visibility or help us use different components. so, I’m a big, you know, I use mostly AWS in my my uh experience. So, we were we were early adopters of like the spot market and RIs and how do you, you know, manage those costs as well as, you know, making sure we got rid of resources or that were wasted or not. So, that’s a little bit and I’ve done that, you know, used I’ve done FinOps or this more of the cloud version of it for the last decade, you know, maybe even before it was really called FinOps. And uh and now, you know, now it’s expanding beyond just the hyperscalers, right? So, uh bills for your observability tools, you know, think Datadog, Dynatrace, New Relic, your data solutions, Snowflake, Databricks, right? Besides the hyperscaler are growing at maybe even a greater rate than what their cloud spend is. So, now organizations are having to deal with all that. And then with AI and what that’s going to do with to your data. So, there’s uh there’s a lot of opportunity in the space to help continue to manage that. And I was just talking to somebody else uh earlier today and it’s it reminds me of going back to the days of CapEx. You have to you have to be more thoughtful in how you’re going to spend your money earlier in the process much like when you’re buying hardware and software. So, I feel like we’re maybe trending a little bit back in that direction.

Itiel Shwartz: What do you think the AI is doing to that space? Like is like CapEx is not important anymore? It just increase over time? Like what what’s your take on like how is the world changing because of the

Josh Schlanger: Yeah, so so pretty broad question. Let me see if I can sort of narrow down a little bit. I think you know AI is going to impact for organizations how they think about their data. Right? How they keep their data maybe near near line versus offline. Right? Cuz they want to use that data for context-related stuff. So, there’s going to be more from that perspective. I also think that, you know, as people use these products and, you know, everybody’s talking about tokens or token usage, right? So, that’s sort of the metric you you know, you determine on how much you’re you’re using. But people are are forgetting also like and I don’t know if you all are using, but you know, using Claude Claude Code.

Itiel Shwartz: Yeah, with our Right? So, that you know, you pay you pay for that, right? Or I pay a lot money for that. Yeah.

Josh Schlanger: Well, it they exactly and then so not only do you have token usage, you’ve got your you know, you got this tool that you’re using Claude Code. And then you also have, you know, maybe others in your organization have a chat license. So, they’re talking to Claude to help them write materials or do some help you some analysis. So, if you think about all of those different things, these providers are going to be a significant I think over time significantly larger part of spend as we move forward. So, so as organizations, you know, continue to use cloud and continue to use all these tools, I can definitely see these AI tools catching up pretty quick as people adopt them.

Itiel Shwartz: Maybe maybe let me ask you like you talk again like on the I understand what you’re saying from the high-level perspective, but in then cost is a lot of the time debate, I will say or at least some tension between the FinOps guys or like the financial people to the developers to the DevOps, right? Like there are like the easy things like use a cheaper machine because you’re under utilizing that. No one will oppose to that, right? Like most of the time. Sure. But a lot of the times the real cost saving is done because we re-architecture our application, right? To we change something in the flow that we’re working or not really. So, maybe like what’s your take on like whose problem is it this like FinOps? Is it the VP R&D? Is it uh operation people? Is no one?

Josh Schlanger: yeah, so so Okay, so the there the I think the execution or the operational component falls into a typical SaaS business will fall in under R&D for sure. They own they own fixing the problem. I do think that they need visibility or awareness of what they what the guardrails are. So, that’s an executive you know, CFO CEO sort of saying, “Hey, look. Here’s here’s the numbers that we need to hit. Here’s what you you know, here’s your forecast. You need to architect and design within these frameworks, right? Based on what we want to go build. We want to have X number of customers. We want to build these products. You should have some I don’t know if you do like architectural decision records or any sort of pre-work as a as a way to determine what you’re going to go build, whether it’s a product, whether it’s uh a new new new product or, you know, adding to an existing product. But you need to work within that those frameworks within those guardrails. But the execution, the delivery of it needs to be the ones who can do the work and that’s the I think the engineers. Mhm. So, uh but they they can’t do it in a vacuum. They need some information. Right? Say say, you know, or guardrails, I’ll call them policies whatever you need something like that that says, “Here’s what you need to go do.” But that my opinion is, you know, for sure engineers need to own that.

Itiel Shwartz: Okay. We are like closing to the end of the time. Maybe you can talk a bit about the biggest like latches or like let’s say I’m a CTO, right? And I’m indeed a CTO. What are the like biggest cost spenders that I don’t even know or like let’s say someone told me I have to reduce my cost. Like how would I like start such a project?

Josh Schlanger: Yeah, so so it’s a good question. What I would tell you is that you need to have some level of visibility on, you know, like, “Okay, here’s where we’re spending, but then here are the areas for opportunities potentially to reduce.” Right? And you need to look at those areas. So, some some big areas if you’re looking at just purely the hyperscalers, right? I would look to see how you’re how you’re managing your and if you’re, you know, this is Commodore, right? So, you’re Kubernetes you know, so I would look at your cloud Kubernetes clusters and are you appropriately sized for the workloads and is that an opportunity? Now, if you have one cluster, maybe not, but if you have lots of clusters, there may be opportunities. But I think the biggest areas for for organizations really at the end of the day is probably going to be in the data space in some way. Whether it’s databases, whether it’s your how you’re managing your storage, your S3 storage and if you’re tearing that storage and moving it off to other other tiers. if you think about uh I’m lumping in the MongoDB, Databricks and Snowflake of the world, right? So, if you’ve got that, you need to think about how how you’re managing that. So, those are the areas that I would focus on. Thinking about data and anything associated with data would be big ones that I would focus on if I was coming to you as a CTO and saying, “Hey, we need to reduce it. Take a look here first.”

Itiel Shwartz: Okay. Okay. That that does make sense. That does make sense. maybe let’s talk a bit about like where we going maybe as an industry. Like what are the biggest trends that you are seeing right now uh in our space. And when you say in our space or you know, FinOps or just more more holistically? Holistic. Well, give me both. Why why not? Like you said there there are like 200 tools in the FinOps space. So, I think like this space is moving fast. So, both.

Josh Schlanger: Yeah. So, I think um specifically for FinOps, I see there’s a couple of couple of trends. I think most organi not most. There are there are there’s this Let me say it this way. There’s a trend towards moving and I talked about a lot about it today, moving to get in front of things. Like what is a governance? What’s your top-down view? How do you get in front of these these uh costs? Cuz if you’re if you’re already spending too much, you know, you’re sort of behind behind the eight ball. So, now companies are trying to figure out what does that look like? When I say companies, these solutions are trying to figure out what that looks like. That’s one. Secondly, you know, and I would say more generally is I’m excited to see how and I’m starting to see a little bit of how AI is being embedded into products. So, you don’t necessarily interact directly with the AI, but AI is is not just from a Claude Code and building the product, but there’s a lot of I’ll call it agents behind the scenes doing different aspects of the work that you don’t really realize, but it surfaces in some way in your application. So, there’s this using really AI as a tool which to help build these companies uh out. So, I think that’s a that’s a general trend that I’m seeing both in the FinOps and I say in more holistically across the board. the other the other interesting thing is, you know, I’ll I’ll mention a really wrong platform teams. Right? So, so I’ll I’ll take it one step back real quick. So, I don’t So, I see that I we use the term DevOps pretty loosely. I think DevOps is much more of a culture than a than a function. Right? And I may be controversial, but, you know, I think, you know, it’s it’s DevOps is about how do you empower engineers to own the full full operational as well as development components. So, who’s going to help them do that? I That’s where I think platform engineering comes into play. They really are the glue that that brings everything together. And I I think there is this sort of trend to slap what we would in the in history called system admins as platform engineers or DevOps engineers. And I think that’s a mistake. I think there’s the there needs to be a proper mindset of what platform engineering is. They’re an enabling and glue team. Right? And when I say enabling, they enable developers to move as quickly and as fast as possible while ensuring that they can operate these systems. So, I don’t know if that’s a trend, but more of my opinion and take. And I do think that we need to continue to see more platform teams really coming about to help development teams uh succeed.

Itiel Shwartz: Okay. That’s quite cool. What is like your prediction of like where are we going maybe as like an industry or like if we’re going to talk in 2 years, what do you think is going to be like completely different?

Josh Schlanger: I mean, I don’t think this is going to be news to anybody, but I you know, how we use AI to do these functions will will is going to be super interesting, right? So, you’re starting to see it with these uh more in the SRE side, you know, AI, you know, for SRE or these automated systems doing, you know, more uh SRE type work. there’s been a there’s a handful of companies out there in the space. So, I would say, you know, and I would actually ask you. I’d be curious to know your opinion, too, here, but I know this is but I think AI is going to be the big one and and how we adopt that in into the different practices and capabilities.

Itiel Shwartz: Yeah, like I think that like AI is the double-edged sword that everyone is both fearing and like using, right? Like it costs us much more money, but it can maybe save us more money. So, I think like that’s the that’s like the biggest like trend that that I can kind of think that like AI for SRE, you talked about it it also a bit like observability and cost. I think they’re also like kind of like starting to merge. as we helping with to the cost optimization, Komodor even where like an SRE because I think those words are kind of interconnected.

Itiel Shwartz: Okay, maybe to conclude like one tip from you to our listeners, like longtime listeners. What do you think?

Josh Schlanger: Yeah, so um one tip. I So, if you’re if you’re struggling or thinking about cost, do the easiest thing possible. Go to go within your cloud providers. They all provide some and you know, trusted advisor like, you know, hey, go do this, go do that. Go do that and and clean up whatever waste and unused resources. If you’re if you’re are new to FinOps or even DevOps, you know, my feedback is go do things. You can’t you can’t train uh and read your way into DevOps. You need to go do and you need to experience. Right? I think, you know, from my experience, all the all the callous, you know, hands that I have, all you know, all you know, all the all the uh rough skin I’ve gotten from, you know, both good and bad experiences, right? You know, and we’ve all been there. But, I wouldn’t have been as good a leader and as good an engineer if I didn’t experience the ups and downs. So, um if you’re considering getting into the space, find a job, do whatever you can, take it and learn. Right? I believe learning, you know, doing the work and on the job is much better than trying to read your way or, you know, certify your way into the role.

Itiel Shwartz: Okay, that’s a good tip. Okay, with that, I think we conclude the meeting. It was a pleasure having you and uh yeah, let’s hope we will save some money.

Josh Schlanger: All right, really appreciate it. Thanks for uh inviting me today.

[Music] Kubernetes for Humans.

This is an AI generated transcript of the conversation