Transcript 0:00 We need to be DevSecOps, we need to do Agile, we need to be ten X developers, we need to do AI prompting, and I also want you to do FinOps. Hey, I'm not saying it needs to be cheap, it needs to be well spent. 0:11 We need, we need to maximize business value out of your investment. Sometimes that means you need to do more scaling, which is totally fine. 0:18 But if you don't know if it is totally fine because nobody is using that little feature that you just built, yeah, then why are you building it? 0:25 Say that you need to spend first before you can make money, that, that's totally fine. 0:29 It is not just a finance thing, and it is also something that you can at least bring to your product teams and say, "This product, how, how much is it making?" 0:37 So I think that story coming down from the top saying, "Hey, we are now driving a good business because our cost is this, but our revenue is this." Welcome everyone to a new episode of the FinOps Weekly Podcast. 0:51 Um, today we have a very special guest, someone that I have been talking since and we have been interacting, so a partner from a newsletter perspective. Today we have Mitchel. How are you doing, man? I'm, I'm doing fine. 1:03 Thank you for, uh, inviting me here. Pleasure. Pleasure to have you. 1:06 And today we have a very interesting talk, especially if you are a developer and you are in the, in the engineering ecosystem, or if you are interested in all these open source, uh, stuff. 1:15 We have someone that has, uh, maintained an, an open source project, and we want to talk about it and how the developer's role is on, on FinOps. Um, but before anything starts, Damian, how are you doing, man? 1:28 I'm doing great. Always a pleasure to be here. And, uh, thank you, uh, Mitchel, for, for coming and joining us. No problem. Pleasure. Pleasure to have you both here. 1:38 And, you know, I wanted to, to start kicking things off for, for, you know, uh, for people to know that, uh, Mitchel runs the, the Azure Cost, uh, CLI and maintains the project, which is an open source CLI tool to, to help you with the Azure costs. 1:54 And, you know, want to ask you a bit more about what drive you to, to create this, what's the benefit for, for a developer to, to use this, and if someone doesn't know the tool, like who do you suggest it should, uh, download it and how to use it, you know, a bit more about how, how this project can help a FinOps, uh, developer or any developer on, on Azure. 2:15 Yeah. Yeah. Let, let me explain some of the background of it because, uh, I'm, I'm a... Yeah. I'm a, a consultant with, uh, with CBR. 2:23 So we had a, a big customer and, uh, they asked the question, "Are we running cost effective?" And I think, uh, you probably recognize the question that it... 2:32 The answer is always no, you're never gonna roll, run cost effective because you always have costs, because you have scaling problems or you have developer productivity, or you didn't make a choice yet. 2:43 So that's an easy answer to give, but they wanted to have the answer on skill, and they had around 200 teams, uh, a lot of, of Azure costs, a lot of different subscriptions. 2:51 So we needed to scan their environment and also scan their Azure costs. And of course, there's a lot of already existing tools in the market, but you need to install those, you need to set them up. 3:02 They start collecting their own database, their own storage. So we thought, "Yeah, let, let, let's make that easier. How can we actually get it from the API?" 3:11 So that is where I started to look at APIs and, um, got, got interested in, and how can you get the data from the Azure environment. And there's a very nice, well, nicely... 3:22 There is an a set of APIs that allows you to get all the cost data. And then I found another project in AWS Land, another CLI, uh, which was pretty neat. It was very easy to start. 3:35 You could just kick it off, and then it would show you the cost of the last week or month, whatever you had, in different- Mm-hmm... outputs. 3:43 And it was really built, and you could run it, have the output, put it in a Slack message, and then do something with it. 3:49 And that really aligned with, with my thinking, what if you could get data out of the system and directly bring it to a developer? 3:55 So you could quickly see, I spent this amount of money yesterday, or this amount of money in the last couple of hours, or this is what we're gonna do in the couple of, uh, the couple of days, the forecasting. 4:07 And combine that with my love of developing, trying to see how we can get the console to do nice stuff, and talking to APIs, I started to build this tool, and it focused on a couple of things. 4:20 Easy install, so you needed to get it without installing all kinds of, um, well, components and build it yourself. 4:27 Now, if you know.NET a little bit, there is, there's a nice.NET tools stack that you can use to get the tools to your local machine. And I wanted to be able to, uh, pick up the authentication you already have. 4:40 And if you're familiar with the AZ CLI editor, the Azure CLI- Mm-hmm... 4:44 that allows you to do all kinds of stuff within, in Azure land, there is, uh, a concept of logging in, uh, you need to authenticate yourself, you need to talk, "I'm, I'm this person, I have access to this subscription." 4:54 Mm-hmm. Let me see if I can leverage that. And luckily you can. In.NET there's a lot of capabilities to capture a context which allows you to call APIs on behalf of the user. 5:06 So combine that with easy install, easy authentication to something else, I was able to fetch the data in all kinds of different structures, like, uh, current cost, cost per day, cost per, per resource, group it, uh, have all the dimensions around it, look at the forecasting cost, and then triggered by what if you can actually format it differently. 5:30 So the console is a very nice way, and it looks very old school MS-DOS, but it, it is a very nice way to visualize stuff as well. 5:39 And I always wanted to play with the, the Spectre console just because I wanted to play with something, and that has very nice bar graphs, and you could make very nice overviews. 5:48 So combined with that, I was able to mimic what was on the Azure cost page and then show quickly to users, "Hey, this is the cost I had the last day or the last week," or, "This is the cost that I'm gonna have."To dive a little bit further, what if you have other formatters that convert it to JSON or Excel- Mm-hmm... 6:09 or Markdown, and with Markdown you could hook it into your GitHub pipeline or any pipeline that you have, so you have an automated process. 6:18 And when I release this tool, of course, uh, I'll just announce it to say, "Hey, yeah, this, this is doing these kind of things." 6:24 And, um, apparently it's got some traction, and I've got a lot of feedback from people that are using it to export and import the data to another system. 6:32 They use it to actually feed it into another tool to actually process the data further. 6:37 They use it to, uh, generate reports, because it gen-generate the Markdown, you can convert to HTML, you can actually send it per email to somebody else and have a, a, a weekly or daily report just by hooking it up to some sort of pipeline that allows you to get the data. 6:53 You can use it to make decisions if the budget is higher than, than this, or you basically have more, um, anomaly, uh, it does some very raw anomaly detection based on your last amount of, uh, cost that you have. 7:08 You can trigger to be an, an, uh... You can trigger to run a negative value, so you can stop your build because you did something that impacted your cost in the past. 7:18 So all these kind of things are being used now by people to hook it into a pipeline model, to generate outputs, to use it for input/output to other systems. 7:26 And I think that is a very nice way to see this tool integrate into an, uh, developer flow in, uh, interacting with data, et cetera. 7:35 And from there it grew to other overviews, getting the price list, see what is the most efficient region to run your, your VM in, what's the most expensive one, the most cheapest one. 7:45 So I'm, I'm experimenting with all the cost data that, uh, gets out of the system. So I'll just give you another overview, uh- That's our overview... of my intention. [laughs] It's, it's, it's very good. It, it seems... 7:58 We see lately in the, you know, in the industry that a lot of the FinOps or costs has been shipped to the, to the de- to the DevOps, to the engineering teams. It seems that you- Yeah... 8:10 like thought about it before, right? Do, do you, [laughs] do you see, also see it in this movement- Yeah... this movement to- I, I, I... 8:17 If you look at the FinOps definition, uh, it, it's a cultural thing, and I think that that's, uh, shared between all kinds of different disciplines. 8:25 Uh, one of them is the engineering teams, it is the developers, it is the people that make the software. And what I see when I talk to developers, it's mostly FinOps. 8:34 You already want us to do DevOps, we need to be DevSecOps, we need to do Agile, we need to be 10X developers, we need to do AI prompting, and now you also want me to do FinOps. Why? What, what is for me to gain there? 8:46 What can I do? Uh, FinOps is something that the finance people do. They do the contracting, they do the reservations, and all the, the saving plan. I don't know anything about it. 8:57 And I think that you have, as a developer, you have a very big stake in it. All the, the, the users optimization is your part. You can do a lot of stuff there if you have the right tools and the right means to act on it. 9:11 And that can be as far as, yeah, the shift left me-methodology. 9:15 If you do it, uh, early in your development process, you can actually make sure you have cost avoidance, that you're not spending the cost because you didn't roll out that resource because you already saw this doesn't make sense. 9:28 Yeah, I'm, I'm scaling it out to 100, that makes, makes a lot of additional cost. Let's not do it. Let's have a budget there that says you shouldn't do this because it has a much impact. 9:38 And I think that kind of help for developers can help them be more active, be more aware, be more, uh, well, working together with the FinOps team to save on the cost and make more efficient choices. 9:52 And then you still have a FinOps te-team that does more of the centralized work, where they talk about, "Hey, how do we educate the people? How do we enforce a culture in, in the company that is 10:02 looking for the best value out of the clouds? Uh, how do you, uh, make, make sure that the right rates are negotiated, that you use the right plans?" Yeah, they need the developers as well. 10:12 You cannot make a saving plan without knowing if this resource is gonna be used for longer term. You need to have that conversation. 10:19 I think the FinOps definition all capture that, but it is, uh, high level and, and I'm trying to bring that down a little bit more to concrete stuff. 10:28 Yeah, you need to, to make it, uh, clear for, for the people to employ. [laughs] Yeah. And I think the developers have, uh, they don't know that yet too much. And of course, there are mature teams which are cost aware. 10:40 They, they really design with cost in mind. They have good conversation with the product owner, how much can this product cost? But a lot of organizations don't. They just spend- Mm-hmm... 10:49 and they just think, "Hey, if I turn this option on, I get the additional features, and let's also turn this on in development, and because that's easier." 10:58 And then you have in development a server running 24/7 with all the bill, uh, the bells and whistles on it and, and yeah, just spending money while you shouldn't be spending that money. So there's no value- Mm-hmm... 11:11 why should you run it? And I think that awareness, yeah, that's, that's for a lot of companies not yet there. And I hope with tools- Yeah... like the CLI tool, it's just a quick way to get started. 11:22 And I know there is also in the ASAT CLI tool, there is cost options. Yeah, but they're very raw. 11:29 You just get an export of your data and, and I want it to be a little bit more graphical about it and show you right in your face you have an anomaly, you can do something with it, or this is in a graphical way the cost that you're gonna make. 11:40 So what are you gonna do about it? I think those things will help developers be more aware of it. Together with all- Yeah... the other stuff, uh, tooling processes, education, training, et cetera, et cetera. 11:51 Especially, uh, and it reminds me of, uh, when, when you do like security validations or validations of some kind where you are, uh, provided of, with tools, new tools that you don't know because it's not like your main-... 12:04 but they help you like, okay, this is not well done. You need to do this, this and that. These are the guidelines, these are the tools that you can use and how you can implement them in your pipeline. 12:14 So now the pipelines, I, I think we discussed this, are larger, like more, have more steps, but the pro- the final product that you get 12:23 once the fi- the pipeline is finished is way more polished and way more finished with security checks, FinOps checks, uh, validations, uh- Yeah... code validation, et cetera, et cetera. 12:37 And, you know, tools like yours installed in a, in a workflow of the, the development makes like all the development process in a single cycle instead of making it different cycle. 12:48 Like you, you deploy, you have a final product, and then everyone- Yeah... attacks the final product, right? [laughs] I, I think we see a lot of companies indeed that, eh. 12:56 You have certainly with finance, it is normally you get the invoice days, days, weeks later, and then it goes to a completely different team. 13:05 It goes sign off by some manager somewhere, and he has no clue what he's signing off on. And I think that is something you want to bring much, much closer to the development teams. That needs good ownership. 13:15 You need to have the right tagging, naming, et cetera, so you know what is yours and what is shared, et cetera. 13:20 And you need to have that shift left methodology and the mindset that you actually, yeah, like, like with security, if you found an, um, um, security vulnerability in production, it's much more costly to fix than when you find it already in development before it goes to all the stages to production. 13:36 And the same with FinOps. If you fi- basically find issues that can appear when you go to production, already fix it here. Yeah, it's cost avoidance. It's the best way you can have it. 13:46 Not spending the cost because you don't need to. Yeah. So I think that is really where you see more vendors stepping in and, and tools to help out. Yeah. 13:54 The preventive, the preventive area instead of the, the reactive area. Yeah. 13:58 And, and I wanted to, to ask you like for, for developers that would be listening to us or, you know, watching, because I know I have, uh, you know, a lot of the audience is from the technical perspective, if you have to say the top two or three actions that a developer can do to help, uh, the cloud cost of what they normally do, uh, be reduced or be controlled, what do you... 14:22 what's your top three, um, actions that a developer could take? 14:26 I, I think and, and that near real time insight in your costs, if you don't have access to that as a developer, I think that's already a, an, an, a red flag. 14:35 And then again, in a lot of companies, developers don't have it. They don't see it. They are being shielded away from it. It is in tools owned by another department that sees all the tools. 14:46 So you need to enable that for your customers. Yes, the Azure portal shows you some of the data, but only if, if you are allowed to, to go there. 14:54 You can see on your scr- subscription level the cost, but not if you are under a CSP, because then the CSP is hiding that cost because of all the overhead he has. Mm-hmm. 15:02 So that's something you need to expose to your customers. That can either be... or so to your developers. So that can either be an, an, its own dashboarding, more access rights, et cetera. 15:10 If you do not see what you're doing, yeah, that, that, that previous effect, if you press the pedal and you see your cost coming up or going down because you reduce the pedal, yeah, they, then you're basically driving blind. 15:23 So that is the first step. I think the other one is, is having cost awareness in your initial conversation with your product owner. A lot of people just start building features, make decisions there, 15:35 and they're not validating if that's a good decision based on cost. They look at performance, they look at scalability. If you're lucky, they look hopefully at security. 15:46 They look at business feature, but not really at the value. And I think that is also an agile thing. 15:51 Yeah, you're, you're really just building what the product owner says, but I think it's, it's good to ask the question, is this something that you're gonna see sold by a lot of people? 15:58 What's the value that you want to get out of it? And if it adds a lot of value, then it is not a problem to spend money, eh. I'm not saying it needs to be cheap, it needs to be well spent. 16:08 You need to maximize the business value. That's the whole definition of FinOps. Not save money, you need to maximize business value out of your investment. 16:16 And sometimes that means you need to do more scaling, which is totally fine. 16:20 But if you don't know if it is totally fine because nobody is using that little feature that you just built, yeah, then why are you building it? Why are you building this with this technology? 16:29 So that conversation needs to go first. And then a very simple one, look at the Azure calculator. It's such a great tool to find out w- how you're being charged for certain stuff. 16:42 And, and a very simple example, a, a service bus in Azure, I think the cheapest variant is, is around a couple of euros a month, and the m- little bit more expensive one is €10. 16:53 And what I see with a lot of companies, they take the one of €600 a month. And why? Because they think, yeah, we need it because, A, it is premium. That could be good because it's premium, and it has a vNet support. 17:05 And we should use vNet support for security. That's, that's... I'm in total favor of that. But do you need this for acceptance? Do you need this for any other environment where you don't process production data? 17:15 So if you don't know the difference, then you're easily spending in e- each environment around six hundred, €700 of additional cost that you don't need to make. 17:25 And same example with, with the App Service, same example with API management. You can make choices, but you need to know what kind of choices you need to make. 17:34 Yeah, so be aware of your costs, have the, the right conversation with the business, and be aware of what makes costs. Where is the cost hidden? 17:43 And sometimes it's as simple as reading the documentation or looking at the Azure portal and, and see the price calculator. I think these three things help, uh, with setting stage for a lot of stuff. 17:55 Those are- And of course the awareness of why you would do this as a developer, what kind of impact you have. Yeah. 18:01 Once you know what you, how, you, what you are spending and how it's going, the money flow, uh, then you are able to do the adjustments, um, and that's it, right? 18:10 LikeFirst you need to be aware, and then you need to-- then you have to develop the knowledge, so maybe that the FinOps, uh, team can help, and then you can do the, the savings, right? Because otherwise- Yeah... 18:22 it's impossible to- And, and it goes in those phases, eh. You, you cannot do this right from the start. No, no, no. You need, you need to learn. No. [laughs] You need to build up. And that's fine, eh. 18:29 Then, then you are at least, you're aware that you're not competent enough, which is also good, say, that you need to spend first before you can make money. That, that's totally fine. 18:38 And at some point you go to a mode where you have more control over your processes and where, yeah, cloud spend becomes more and more important, and then you need to get into those processes. 18:48 Sometimes that means tools, sometimes it means guidance, guardrails, eh, like, like this is where you can go to within that boundary. That's totally fine. 18:55 But if it goes over the boundary, yeah, maybe you should get an alert of a warning, or you should get an email or, uh, maybe even stop the build because it goes over a certain set of, of, uh, of money. 19:06 I think these processes will help you to still have, uh, speed in your process, because I don't want it to be limited. I don't like a FinOps team that are saying no. That, that shouldn't be the case. 19:18 Uh, you want a little bit of control there. That's, that's totally true. And swapping a bit the, the... Following up, but swapping a bit the, the orientation, making it on, on the C flat part, like being a... 19:32 Because I like the, the orientation of, of making pre, or let's say, proactive approach. 19:38 So basically what I'm saying is here, like, uh, from someone that is a cloud architect or a developer, uh, designing a project that is going to be on cloud and, you know, for example, a cloud migration or a new product. 19:52 Let's say, what are your, your advices to make the cost like a, a requirement, and what are your advice as well to build the cloud architecture in correlation to the cost? Because now we're seeing like... 20:05 Or for example, when I was designing, cost was not, was not a factor. It was efficiency, specific technologies, uh, the tech stack, uh, all this stuff, but not the cost. 20:15 Like the cost wasn't considered up until the end when you see the bill, right? Yeah. Yeah. 20:19 But for someone that is designing, uh, what are your, uh, advices for making a cost-efficient, uh, architecture for, for a product? Yeah. 20:29 I, I think here it goes indeed back to, to that question to your business, what is the goal? 20:35 If the goal is, is going quick to market, yeah, that's something you need to st- you need to bypass some of the process and focus more on the technical point of view, that it is secure enough, performant enough, and you can quickly go to market. 20:46 If it is about, hey, this is a solution we know they're gonna run for a couple of years, yeah, maybe it's more efficient to look at cost elements as well. 20:54 And if you build a system that is, let's say, uh, heavy data-driven, so you do a lot of calls to your system, yeah, then you need to consider data calls, yeah, also cost money. 21:05 Would it be more efficient to cache data or be more, um, um, more strict on the CDN or what kind of, of stuff to basically offload the calls there? So there you make other kinds of choices. 21:16 Then a CDN is not to make sure you have low latency. The CDN is there to offload your more expensive backend server or your database, and you don't want the calls to happen. 21:25 If you choose a more traditional VM-like approach, yeah, then you need to look at, uh, do I need to scale it down at night or does it make sense to run twenty-four hours? Do I need to, 21:36 uh, scale it up, scale it down in another pattern because that is more efficient? 21:40 And if you don't ask those questions, you just let, let it run for the whole day, yeah, that can also be a good answer because you don't want to spend time now on that problem. 21:49 So again, it goes back in, in collect those requirements. And, and if you design with the cost in mind from a more architectural point of view, you should have that conversation. And it can be a simple, "We don't know." 22:00 That is also an answer. And, "We don't know yet," is also a good answer, but at least you have started that conversation, and I can still make a decision later. 22:08 And I think you get a different outcome if you look at cost than if you look at pure performance or scalability. Right. I think, uh, you got a point there, I think. Uh, and, and of course there need to be... 22:20 The FinOps needs to go there and, and ask the right questions, right? You have the need to s- in, in the design, there need to be like a setting which the FinOps steps in and just ask questions. Yeah. 22:31 And, and I think it also helps, yeah. The cloud is there, uh, is, is not a platform to just run compute on, yeah, because it's not just virtual machines. 22:40 There's a lot of services that, that if you select them rightly, like, like the serverless functions, make a very good use case for a lot of functionality. 22:49 But for other stuff it doesn't make a lot of sense 'cause of, of the, the time it could take or the, the scalability that you might want to achieve or not. 22:58 If you don't make those choices, you, you, you will not end it, end up with a good architecture. 23:03 So cost is a part of it, but I think, uh, and the thing we always use it as a little bit of an, of an additional attention point in your mind when you pick the right PaaS services or SaaS service in the cloud. 23:15 So it, it's always needed. And I think the cloud offers you all those capabilities. I think a lot of developers might not look at it enough yet, yeah. 23:23 They just pick another criteria like performance features, scalability that you might have, not directly on cost, but it is an important element to consider, and you... I think you get a better application. 23:36 I think that, uh, yeah, the, the developers don't look at it because, uh, I think that also management needs to come in here and put them in their performance review. It also kind of correlates to, to the- Yeah... 23:47 previous, the previous questions that I, I wanted to ask. [laughs] Yeah, yeah. Well, you know- You know, the... One thing, you, you, you don't need to blame also the developers. Also, where are the managers here? 23:56 If you're a- No, but you, but you can turn it around, eh? Yeah. Exactly. If you want to do it from top down, you, you can promote or at least make sure everybody's aware of it, so you, you, you can drill down. 24:07 Cost is important also for me as an organization. It is not just a finance thing, and it is also something that you can at least bring to your product teams and say, "This product, how, how much is it making?" 24:17 Because a lot of teams don't know how much the product is even making in, in revenueAnd I only hear at some point, yeah, this is a very expensive if you look at the cloud bill. The cloud bill is one thing. 24:27 You also have your, your people working on it. And I even have customers saying, "Yeah, maybe we should go back to the traditional data center because it is a lot cheaper." 24:35 Yes, and traditional data center can be cheaper, but you need to do upfront commits, you need to buy stuff, and you need to have the skill set to maintain it, and I think a lot of people forget about it. 24:47 So if you don't take all these things into consideration, uh, the cloud is in a lot of times cheaper, more efficient, more efficient to run if you do it correctly, if you use the cloud's elements of it. 24:58 And I think if you don't let-- if you don't from a management point of view show you that, so this is a project, costs us... 25:04 We have five people on this project, but the, all those five developers are actually developing, they are not maintaining. They use the right stuff to actually add value, and they're not just maintaining servers. 25:15 Because that's what you have when you go to your data center. Yeah, the backups, the disk replacement, all the other things that you don't want to do anymore. 25:22 So I think that story coming down from the top saying, "Hey, we are now driving a good business because our cost is this, but our revenue is this," 25:31 I think that is a more efficient picture than saying you need to be more... you need to be cheaper, you need to be more cost effective. Th-that is a good point- It doesn't work... 25:40 th-th-that they need to start talking about business value. Does it make sense from business- Yeah... perspective? Yeah. 25:47 I think you want to maximize the business value out of your cloud, and that the cloud offers you a lot of tools for that. Th-the scaling up, scaling down. If you turn it off, you don't appear in the cost. 25:56 So there's a lot of features there that helps you, but you need to use it, and you need to make use of it correctly. And sometimes it doesn't make sense to scale because, yeah, you want to go fast. 26:05 You need to get quick feedback. You need to get a server up and running. It takes more development time to build that, and you don't have that yet, which is okay, as long as you're aware. 26:15 And then you have to discuss- Mm-hmm... with your product owner, and then you have to have discussion with your management team. 26:20 Say, "Yeah, we are aware this costs now a lot of money, but, yeah, look, look what we're gonna bring in for the next three months, and then we're gonna talk about scaling down," which is also fine. In it. 26:30 And if you- But you have to have that conversation. I-if you... 26:33 And also I, I, I noticed that personally that when you are engaged, let's say, in the conversation of the KPI, the business KPIs and what, when you know what your product brings to the company or to the customer, uh, not like at this quarter report will say, "Yeah, send me a statistics or whatever." 26:50 It's not like when you know, like you can go and check how much, you know, users you have per month or how many- Yeah... documents you have processed. You... 26:57 Once you know, the developer knows that, it's much more engaged with the product. Yeah. Um, and then it can relate, okay, are we scaling good? Are we scaling bad? 27:07 We are spending more, but the number of documents, let's say, is the same, so we are not probably scaling good. 27:13 We are just, uh, spending resources because we have installed whatever update that doesn't work, uh, anymore. Yeah. But we have, you know, r-re... 27:21 right size wrong the, the resources for our Kubernetes cluster, and we are overspending because of whatsoever. I think once you relate that, 27:29 the developers are much more engaged and much more prone to take action and know the business value that they do. Yeah, indeed. 27:36 And I think that, that cloud unit economics, yeah, where, where you look at how much is an actual user costing me. If I onboard 10 additional users, is my cost linear going up, or is it exponential going up, or is that 27:49 not really impacting because I have chosen an architecture model that allows me to scale the cost over multiple users, eh? 27:57 So the more users I have, the cheaper I am, because that makes, makes a lot more sense in the end. But if you pick a model that's heavily focused on security, then every user gets his own, own, own environment, yeah. 28:08 And that is of course a l- a linear growth with what you use. You're not gonna make, make a huge profit on your cloud setup in that sense. 28:16 So these kind of things, if you don't consider those, yeah, you make other architecture choices, and you... 28:21 I think cloud unit economics is, is a very tricky subject, but if you do it right, it can really help you to have that conversations. Is the product really successful? And it's the core. 28:33 Like, it's the core of the business. It's the core of what it needs to be. 28:35 I think it's been a forgotten ecosystem, and I think where, it's where finance can have more, like an educative part as well, and can be mu-much more involved in the, in the development because they are very... 28:47 they are much more aware of what- Yeah... the to- the, the whole thing means, and that's where the FinOps, let's say, finance part can be like, okay, how much are you spending then? 28:59 Uh, this is the overall bill, then, okay, which part is the specific cloud, uh- Yeah... or the specific, uh, Kubernetes deployment or specific that. 29:08 And then from making, being able to, the developer to digest or divide the bill and the finance to relate to the business- Yes. Yeah... this is where you build the, the unit economics. 29:20 That, that it is very interesting one, because it's very hard for developers to understand what is actually being sold at the other side of the company and, and how that relates back to your financial processes, and then how it relates back to technical processes. 29:32 So you do normally have some, some sort of nested cloud economics where your, your top management reports over different units than your development team probably reports on. But they need to align somewhere. 29:43 They need to be, have some, some clue. Then I think the FinOps team, the centralized FinOps team is the one that can help you, yeah, make those links together. 29:51 So you see this is costing money because this is more optimized or less optimized, and you can link them together. But this is really tricky. Yeah. 29:58 I don't see a lot of companies doing it, doing it enough, doing it efficiently enough. 30:03 They look at the other kinds of costs, uh, mostly on, on project costs, how much, say, with five developers, they spent a couple of sprints, so that is 50K, that is 100K, because this is the amount of people I need to set up for this project. 30:17 And that is nice, but at some point it goes into a mode that it's constantly run. Yeah, what is then your running cost? Yeah, what do you really spend per user, per, per hour, per minute, whatever that it runs? 30:27 I think a lot of people don't know that. Well, this is a very interesting question. 30:31 [laughs] So, and, and I wante- I wanted to, to drive back, uh, the topic, uh, a bit and, you know, since you, uh, are also writing stuff and you, you know, you have your Azure FinOps, uh, newsletter and, uh- Yeah... 30:45 the MindByte, uh, newsletter. And you have the- the- the- the open, uh, which we'll leave the links in the description along with, you know, FinOps Weekly and everything. 30:55 Uh, but wanted to ask you a bit more on the open source perspective and for, from the, let's say from the contributor perspective, because we may have, uh, people, uh, listening to us that, you know, want to learn, want to contribute. 31:08 How this contributing mode, let's say, that is also something that I do- Yeah... how is it helping you, and how has evolved, let's say, your career in FinOps and as a professional? 31:20 Because, you know, I always, you know, say yes to people, okay, there is a form in the newsletter if you want to write a blog. We'll send- Yeah... with you and we'll ask you. 31:29 How this has helped you, so you can inspire other people listening to us to do the same. Yeah, yeah, yeah. I, I think, uh, yeah, for me, sharing is caring. 31:37 I, I think it really help to share knowledge and, and indeed I do that via newsletters, my open source tools. It's for me a way to get my thoughts on paper and, and have some sort of, of reference as well, yeah. 31:49 So I, I like my, I like to put my own ideas, so write them down and I, I also share it. I share it on LinkedIn, I share it on, on the, uh, on, on paper by having the emails, et cetera. 32:01 Um, but the interaction that- that's the interesting part, and I see mostly on LinkedIn I get a lot of, of feedback from, mostly from vendors which want to sell another platform. 32:10 And not like to sell to me because they see I'm a consultant, so it's not very efficient to sell me a cloud platform. 32:16 But mostly this is a service that, that, that adds value if you sell basically advices to, to your customers. 32:23 But for me it's interesting to see what kind of problems they're tackling, what kind of things in the industry are, are really working well or not. So for me, that is already a validation. 32:32 The story I just said about shift left, a couple of companies which are really into that field, while in the past you saw companies building dashboards. That was the thing. 32:41 You, you ingested all kinds of data for all kinds of cloud platforms. You build big dashboards, and then you had a couple of people that looked at the dashboards and did something with it. 32:50 And what I now see is, is more focused companies that are either on the centralized part, yeah, on the savings part, and you have companies that are more on, how do I get the developers to do something? 33:01 Uh, to getting, uh, alerts directly on their screen and say, "Hey, this is something you need to fix. I can fix it for you. I can do it automatically if you don't respond even, but, um, we can do it for you." 33:12 Or even further, already show you the cost inside your Visual Studio. Uh, and, and when you do a pull request, I already say, [mouse clicking] "Hey, this is gonna change your cost from X to I. 33:23 [bell dinging] Are you aware of that?" Uh, these kind of tools, and, and in this case I'm mentioning InfraCost because I really like the tool to, to help in that shift left process. 33:34 They, they take a niche in the whole FinOps, which basically validates what I'm also trying to advocate. So for me, it's really nice to see people around it and, and have that conversation. 33:43 And, and if you look at more the open source part, 33:47 I get some nice books in, because the cost APIs in Azure are tricky, and I don't have access to all the 30, 40 different subscription levels and, and not even to all the same amount of, of cost that some of the customers have. 34:03 Uh, I got some nice book reports that, hey, we try to, uh, render forty million in the system, but, uh, you, you, you don't have room enough in your screen, so I needed to get some, uh, some formatters in there. 34:14 Yeah, that's not normally a value I test with in, in some of my subscriptions. And you see that Microsoft is moving, uh, to other kinds of subscriptions levels, other kinds of APIs. 34:23 So that kind of feedback is very nice from people in, in the field, even from Microsoft itself. 34:28 I'm, I'm an Azure MVP, so I got some, some links to the, to the cost people, um, got some in- feedback from them about what we can do, how we can set up stuff. 34:37 So I think it's, it's a very nice industry that is, uh, a little bit niche, a little bit small, a little bit growing as well in a lot of factors, yeah, and, and getting more and more value. 34:48 And, and I think it's also about time because we sold DevOps for the, um, consulted for the last twenty years. We sold DevOps, SecOps, and now we're, yeah, we see, hey, cloud is there. It's clouds that stay. 35:00 We know how to do DevOps. How do we get the cost under control? And now FinOps is a, is not a replacement, I think it's an extension to the processes. That, that helps this conversation. Uh, that's very good. 35:14 I think that also, uh, you know, to all the content creators out there, you know, it's... There is a saying that until you don't teach, you don't know. So when you share something, it's like you are teaching. 35:26 It's like, like you said before, you know, you are thinking about it. You make it clearer- Yeah... because you want to express it, and I think it's a win-win. You also... 35:34 It also help you with, uh, with, you know, connect with other people and share ideas. So- Yeah... again, it's a win-win. That's true. Yeah. You understand it more, and you share it with them, and you gain... 35:43 It's a win-win for everyone. Uh, and you really dive into those concepts and you look up the documentation, play with a couple of, of things, and then, uh, yeah, um, also from your own experience. 35:53 Uh, I'm, I'm a more broad consultant, so I'm, I'm currently doing AWS jobs, which is completely different than how Azure is doing stuff, which is also nice for the comparison between the, the two, uh, systems and how they do cost management. 36:05 So you learn on, on the job. You learn by, by experimenting. You learn by, uh, by reading and, and investigating. 36:12 Yeah, and then writing it down is just another dimension of how do you get your thoughts on paper and, and share it and get feedback. Yeah, it's, it's... I like it. 36:21 I think it really helps to improve yourself and let other people also learn along, get more knowledge. Yeah. I like it. Or as a recommenda- 36:31 a-as a recommendation from, from someone that, uh, uh, does it, you, and has, you know, previously been stuck in the learning process that you always learn at, at-Uh, as up until some point, and then when you start contributing, you get to a whole another level, and when you start teaching, you get to a whole- Yeah... 36:48 another level that because you are-- you reflect yourself that you do- maybe you don't know as much as you think you know. Yeah. But it's also good because this will help you learn better and learn more. 37:02 Uh, and this way, uh, that's why I always suggest people to, you know, write, do exercises, um, everything practical because learning, you know, seeing material is okay, and it always includes- Yeah... 37:13 but it's the first step. Then it's putting yourself out there and, and sharing the content with, with the others so, so you can improve. 37:19 Because otherwise, if you don't share it, you don't get feedback, and without feedback, you can't improve. [chuckles] Yeah, and, and that's totally true. 37:26 I think a lot of people m- underestimate how many people are actually sharing. It is a very small percentage. A lot of people are, are just consuming, which is fine, eh. 37:35 I'm not saying you need, you need to produce, but it helps you to be... it needs more, uh, learning more stuff. 37:42 It helps you to be more active, and it also learns how you connect to other people and get, get more interaction like, like we're having now. 37:49 I probably would have never met you in this capacity if I didn't have the conversations we had in the past. 37:54 So it, it helps to broaden your horizon, and I, I always welcome people to, to be m- more outspoken, to share the information they have, and I hope it's an open industry and open world where you can just voice your opinion there. 38:09 Yeah. Try, try to build something. You know, I think that always helps. Yeah. 38:16 And if you think there is always people, you know, uh, listening and will, uh, in especially in this community as normally, you know, Damian said that he's very surprised that, um, there is a lot of contribution, there is a lot of well thought about, there is a lot of- Yeah... 38:30 uh, benefit of, you know, people looking forward to it, l- eager to learn. There is, there is not as much content as there is in other topics, so I think people- Yeah... 38:41 is eager to learn, and it's also eager to help getting you better on, on contributing, right? That's, that's something that I also feel when, when I see that people are surprised, like, "Yeah, I didn't do very much." 38:52 Like, you are probably one of the unique- Yeah... people that contribute, so please continue doing that because I, I... 38:58 it's, it is super difficult for me to find this type of content, so I always like encourage people to, to do the same. [chuckles] Yeah. And, and it is broad. 39:05 FinOps is, is, uh, I'm not saying it's new, but it is, is relatively new. It is based on, on a technology that is still growing, and like the cloud is, is constantly changing as well. 39:16 And it consists of a lot of broad subjects, yeah, from, from developers to finance, to business, to unit economics, to working together with DevOps, but also, uh, AI. 39:27 How do you use AI in, in Dev-- in, in FinOps or the other way around, what is FinOps in AI? 39:32 So there's a lot of things you, you need to consider, and I really love to read stories about how people do it in their organizations and how, what kind of benefit they got out of it, how much they, they optimized or saved because they, they looked to find something. 39:45 Yeah. There's some good stories there that they go, "Okay, I never have thought about that," but you apparently you, you created the right report. 39:52 If you looked at something or going from version X to version I already helped you save this amount of money or choosing another architecture. 40:00 Those real life stories, I think that those for me are really valuable because they prove that, that it can really work, it can make a difference. That's what I really would like to see. Indeed. 40:10 That, that's super inspiring. Uh, so Michel, pleasure to, to have you today. It's been a wonderful talk. 40:17 I hope people have learned a lot about the tool, the work that you do, and we left all the links in the, in the description below. It's a pleasure to have you. 40:25 It's a pleasure to have you in the community and also to, to, to work with you and to, you know, share stuff, uh, together. So man, pleasure and- Yeah. Thanks for having me... thanks for, thanks for attending today. 40:37 It was nice, yeah. Thank you very much. [chuckles] Thank you, and see you everyone in the next episode. Bye-bye. Thank you. Bye-bye. 40:47 [outro music]