Cloud Native Computing Foundation KubeCon + CloudNativeCon North America 2022, 12 Nov 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: FinKube – Making the Business Case For Kubernetes At Your Company - Somik Behera, CloudNatix

Description

Don’t miss out! Join us at our upcoming event: KubeCon + CloudNativeCon Europe 2023 in Amsterdam, The Netherlands from April 17-21. Learn more at https://kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects.

FinKube – Making the Business Case For Kubernetes At Your Company - Somik Behera, CloudNatix

Speakers: Somik Behera
This session is for DevOps leads and managers that are driving Kubernetes (K8s) adoption within their company and want to level up their influence and skill set in making the financial business case for adoption of K8s and associated CNCF ecosystem components. As Kubernetes crosses the “chasm” in the technology adoption lifecycle, K8s champions now have the opportunity to further drive K8s adoption and transformation of their enterprise. This new phase will require “T-shirts” (Dev & Ops teams) to speak the language of the “Suits” (Finance and execs), figuratively and literally. This session will provide you the toolkit to make that business case, capture results and demonstrate value from K8s roll-out and expansion across 4 dimensions: - Capacity & Cost Optimization - quantify the bottom line benefits - DevOps Productivity - quantify the benefits in % of FTE savings - Developer Velocity - quantify the reduction in time to value - Business agility - finally, quantify the business impact of new K8s based services on top line We will open-source and provide the Excel, Google Sheet, Google Slides and Powerpoint templates used in this talk to help the community create a compelling business case proposal with solid ROI.

A

So, let's get started uh good afternoon kubecon, you know: I'm Sonic Bera, here I'm, head of product at a company called cloud natics spent last 20 years in the armpit of infrastructure, as we know it in the end, in the open source space, starting with being an early engineer in virtualization at VMware to creating software-defined networking, the company called nyseria create an open source project called Neutron, which was a subsystem of networking for openstack and then create the data center operating system as commercial product based on the Apache mesos project.

A

But today I'm not here to talk about technology, but something rather more dear to my heart, based on my experience on these years. Working in you know technology as well as more recently in the cloud native space, and that is how do you make the business case for the technology? How do you make it cross the chasm? How do you actually Push It Forward in your organizations right?

A

So you know how can we help justify the value that we bring to the table, not just Implement? And finally, how do we measure the value we are providing after we Implement and close the loop close, the feedback loop? We we did what we will say we're going to do and we showed that we did it and then finally close that Loop right. How do you do that and I believe?

A

As Cloud native crosses the chasm, it becomes imperative from the community on all of us that we enable the t-shirts to speak to the suits right, the devops folks to be able to speak to the finance and the line of business Executives to actually make people understand why we do this and I hope today, we I can give you one of the tools in your toolkit that will help you do just that.

A

So you start with kind of the agenda. Why, then, the how part of it and then what we'll do a demo of what this toolkit is and how you can use it and how we can actually participate and all make it better to be hazard right.

A

So let me start with something um wife and Cube, because there's a code that uh struck a car with me is uh that says: no one will pay you what you're worth right?

A

No one will ever pay you what you're worth they will only pay you what they think you're worth or your project is worth uh and the good news here in that code is uh you or we can control that thinking? If you, if you, if you set it up the contacts uh correctly, you have that control. If you don't, then you have it's a lost opportunity right um so who wants to get paid more over here me there you go lots of us.

A

We can take a lot of these lessons and apply to our regular lives, I'm going to focus really on cloud native and what it means to control the thinking on actually moving on a cloud native project. If you have a non-cloud native non-cubernetes infrastructure, how do you make a case for that? But the some of the Primitives are some of the concepts we'll talk about applies everywhere. So there's something in it for everyone.

A

You know if you're, a technologist you're an engineer or not uh so the way you control that thinking um is by clearly defining and communicating the value you have defined the rules of the game. This is this is why why it is important how it's important a priori and then communicate properly so that we are aligned on what are we doing and how we're driving that value?

A

And if we don't do that right, we want to see more of this. This uh two articles that showed up in last week or just looking, it's like yeah, we've done all this work.

A

um It's not useful effectively, and that is a kind of a that's where the objective is misguided, because you know some projects, some companies, please didn't Define their objectives and didn't communicate the value people say it's not worthwhile, it's it's or or worse, yet you've got to go on a failure mode for the project.

B

A

So that's the high level objective. We want to clearly Define and communicate value taking the next step. So how do we do that? For things we care about in this community is to help kubernetes and Cloud native in general, kubernetes and specific cross. The chasm and become kind of standardized infrastructure, layer and I'll start that, with with the journey, I I hope our users take.

A

We have a set of tools in in the community. A lot of folks. You have folks here have heard of the landscape project or landscape.cnc file. It's great, it's good for Discovery.

A

You go and discover the specific tools you need in your toolkit to implement a specific use case initiative or a project, and then we jump directly to implementing those right to uh getting dirty with your terraform or whatever installable you have, and then the yaml files for modernizing the applications Etc um and then, of course, the once you have implemented it all the challenges which most of the folks in this the solutions exchange and the sessions here talk about is how do you operate and optimize and keep the lights running right?

A

um We keep. We keep jumping the loop, but what we miss is that opportunity in between discovering and implementing, which is justifying and defining what are we doing? Why we're doing how we will measure success so that when we actually operate and optimize everywhere, you get the credit that you deserve for what you did uh and that's what I hope. Finn Cube fills the gap and to better understand this toolkit I'll talk about you need to have.

A

We need to talk about a little bit basics of pricing and how people think of value and uh how they communicate right. It's this is a rough hierarchy, kind of I framed it as a you know, in the order of Maslow's hierarchy of pricing, if you will right the base layer starts with cost. You add up all the cost for your kubernetes project, people, Hardware Cloud, um you know time, add that cost and that's that's the easiest way to say this is what it costs.

A

That's, how people communicate value, which is not the right way or necessarily but the East. It is the easiest way because you can do math. A lot of us are Engineers. It's easy, and this is what it costs um and if somebody's a business selling that they will say that okay, we'll add a margin on top of that cost and that's our business.

A

Unfortunately, internal companies, internal teams, Tech Ops teams, devops teams, it Ops teams just out of the cost, and then people view as not a value Creator, but as a cost center that needs to be reduced and optimized and not as a value Creator because of the model, how we can Define and communicate our value again right.

A

So the easiest way to take the next step would we think of a value-based pricing or value of the kubernetes project that you're bringing on bringing to your company the different kind of objectives and I'll talk about that in the toolkit.

A

um Specifically, if you look at the value-based pricing, it talks about not the cost, but what it is doing for the for the company is how it's making it efficient, how it's making developers more productive. So if you can release more software faster, what does it mean to your business? Is Top Line? How is it making the infrastructure more available? So, typically, you know a single outage costs about three hundred thousand dollars an hour on an average company may be different for your company.

A

So if you shave off ten percent of that, that's thirty thousand dollars an hour multiply with number of outages. You have there. You have another Vector, the kind of availability vector and that that's monetized right. That would be a way to communicate the value that you're driving versus the cost, you're optimizing and really our My Hope Is as you as folks start using that terminology and go from just being a cost center to a value Creator.

A

We can move to the next step, which I call called true price, meaning not only the value you're creating for your company, but all the impact you're having based on that, how we're having the sustainability impact, because things are more efficient. uh What is the environmental impact and what is it really a true price of that initiative? Not only price in terms of dollars not only for a price in terms of value but in price in terms of the whole of the environment and everybody else that participants right.

A

So that's where we would like to go, but where we are is very much a cost. Center cost focused area, uh crawl, walk, run approach would be let's what if we can just move to the next step and talk in terms of the value we provide in front of instead of the amount we cost.

C

A

With that, let me introduce to you the the toolkit. You can go to fincube.io and get a lot of the resources, the GitHub repositories, everything else today. It consists of three things and I'll demo that to you once uh today's discipline uh kind of an Roi calculator, it's an Excel sheet with, like eight different sheets, um very mvp-ish, so we can get started and expose people to the idea and see how folks use and what the feedback is.

A

Once you do that you get your Roi for uh for one use case, which is moving from a non-cubernetes non-cloud native infrastructure to a kubernetized cloud native application architecture, and based on that it will says what is what is the potential perceived impact for certain size clusters, for infrastructure, efficiency, for developer productivity and for infrastructure or application availability right, and it will have the hypothesis or assumptions I have made, but because it's open source you can customize that model which you know which may benefit your company.

A

If you don't agree or doesn't work, the standard model doesn't work for you out pops, like the graphs and charts you can. You can show to make that business case based on the three objectives. I mentioned out: efficiency, developer productivity infrastructure, available, okay right and then you have a very poor man's toolkit or Poor Man's, a way of very quick and dirtily, making the business case of why you want to move forward with this project.

A

um So that's the first two lags the third leg will be a set of test cases or things you want to accomplish to prove out. So, first to our hypothesis like this is how much our infrastructure efficiency will increase. This is how much a developer will be more productive. This is how much my outfit is will go down. It was like great now I'm doing the project. Maybe I did a pilot or a POC or a small scale. What are the things we are measuring?

A

What are the metrics, we're measuring and that's how we put in our you know, test plan or POC plan and and see you know before kubernetes and after kubernetes. What is the mean time to resolution or what are the number of bills, or uh you know, release a ship to per quarter per day, and then you can kind of close that Loop in a very quantitative in a very logical way, because once you have the numbers, it's very very clear what was before and what is after and what actually got the other word.

A

And sometimes you know mistakes happen. You make an hypothesis and it'll probably may not be 100 right, but at least you have the framework to now continuously improve that model.

A

So without further ado, let's head over to give me a second.

A

To fincube.io and uh mirror my displays and I can give you a live demo of the toolkit.

C

A

You can go to the fincube.io. The first thing you know you can subscribe for updates to toolkit. You can join the slack channel to give feedback or go to GitHub. If you want to kind of make changes or file issues, but I'll just go to the resources tab which will open up a kind of a shared Google. Drive I have the license files in there. As you can have Apache too. You guys can use it as you wish. I modify it.

A

The first piece is the ROI calculator that I mentioned. It's got Instructions make a copy for yourself.

A

The first tab of this is pretty simple: there's once the columns that are highlighted are all you need to fill out now, all the other knobs you can. You know it's like the advanced user mode and hopefully over time, I make more kind of a blog post on how to use that. You can talk about your project name use case. Today we have only one use case which is moving to kubernetes from a non-kubernetes infrastructure future. We can talk about automation, application, delivery.

A

You know more Cloud native application, Services other use cases as well talk about how large of your project you're talking, tiny 10, node cluster. You know small 100, note medium node Etc, let's thousand node cluster, and you can talk about. You know what size of VMS or nodes you're, using and uh without kubernetes, and our assumption is that, because of the intelligent cluster manager, we should be able and normally assumption we have seen it in the industry- is that you can use larger nodes and pinpak more services in lesser number of nodes.

A

So you can talk about you know what are your ideal size of kubernetes nodes? You will go to from your non-communitized node and expected your current expected utilization. You can look up at your actual numbers, so you know whatever it is, and nine percent what I have while it looks low, it's very much with an industry standard.

A

If you go into any environment, you see maybe 20 15 20 on some companies, but really it's in that and you can make the Assumption you'll increase it increases by 2x or 3x, and this is based on real life experience. What I've seen in some of the studies folks have done. This is about two to three x: utilized net utilization increase, you see, forget, requests and limits, but really the raw utilization number.

A

If you look at the operating system level and here you're making it goes from nine percent to 27 percent right uh everything else, you can see how it computes on how many developers devops salaries, developer salaries, Etc customizable, um and that's it and the rest. The next step is you go to the cover sheet and it gives you.

A

The graphs to you know, have your first point of view of your straw, man on a business case on operational efficiency, developer productivity and kind of application availability, um how much you're going to save before kubernetes and after kubernetes.

A

So once you have this, if you want to customize it, of course you can go to all these tabs which have details for each one of this value drivers. If you will right, so you can customize all the assumptions I made around before kubernetes and after kubernetes, but that's kind of the advanced mode player mode. If you will so for most folks, this should be sufficient. You have the initial the use case.

A

We will take questions at the end and then you go back to the finq project, the business case, and it's got a template kind of a presentation where, based on the numbers, your environment, you've calculated you've put in for each one of the objectives.

A

What do you expect the return or the impact, building right and again paste in the graphs and make why you think those will you know sometimes these are the points of Y will make that impact will be custom to your company or your application, or how much how critical downtime is or what is the real true cost of your downtime Etc right and uh once you have that you can.

A

You know, show the summary of your total kubernetes project, your operational efficiency impact, develop a productivity impact and your reliability or application availability outage reduction impact, and that should give you, if not everything, a very good, strong foundational straw man to make that business case internally.

A

Right of why you're going to take on that kubernetes use case that.

A

And I promise we'll have I think we'll have plenty of time for qma or a little time at least once you, and with that you know that is kind of the foundation of defend Cube. You guys can go ahead, try it pretty low bar to entry. You know, if you want, you can create macros and other things, and uh but for most folks they can just get started.

A

It's like Excel and PowerPoint and uh and the reason I started with an MVP of an Excel and PowerPoint is because the the roadmap or the thinking is that what is going to think and how people use and which pieces are confusing and which is valuable. Imagine it being a V1 beta, spend six to nine months, uh get that feedback and you know see which things should be kind of pre-populated or which should be in like the easy player mode, which should be in advanced player mode to get that feedback and then make it.

A

You know, call it GA and expand to more use. Cases like what is a uh can I can I do a similar. You know TurnKey easy business case for application delivery or for some other kind of use, cases which is additional value, Beyond moving to kubernetes that it can do, and once we have that kind of the bootstrap. We can, of course make it like a web app.

A

um That's the way. Landscape.Cnc.Io is based on community interest. How many people want to do it? Who wants to work on it? Etc, it's probably easier. Once we have this working in just a non-code or Excel code working, we can move it to like a web app code, so people can self-serve and becomes a easy use. Business casing toolkit not only for kubernetes. A lot of these principles are Beyond kubernetes.

A

So if you're starting an open source project or you're starting a company or product, you can use these same principles and should be able to build up a pricing plan of how you want to show why you are thing is worth something or what is it? What is the impact? That's happening right.

A

So that's the thinking on the roadmap.

A

So, to quickly summarize what we went through, you know, you're, clearly defined and communicate value. To make your project successful. You need to intercept the journey early and do that, build that definition early as early in the journey as possible to to ensure that you set up the company set up the project for success.

A

Second, should join the fincube community, try it out give feedback, use it and that's how you know you will provide value to your company and to the community, and we can Upstream this back to the masses that everybody else can leverage that selective experiences of each one of us so that we can actually make technology the cloud native movement a lot more mainstream as we cross the chasm as we go into larger companies and bigger initiatives.

A

um And, finally, that's how you become a hero and you know, get promoted right. So uh that's that's. How I had I had a top level uh I think we have plenty of time to uh take questions uh so I'll go ahead and open up the floor uh for questions and for once the questions we don't cover. I will be at Booth su-84 uh on the showcase on the show floor, Cloud natick's boot, to answer from there foreign.

D

Thank you, um I, really like the way that you quantify uh some of the advantages with kubernetes on a like a numbers based level. I think that's really important.

D

um I was wondering if you could speak to the calculation of the value for um improved developer. Velocity I saw you had on one side um and then a minor point of suggestion. I believe the graphs for costs with and without kubernetes. The Legends are reversed on the first two graphs, which makes it seem like kubernetes increases, cost yeah.

A

Yeah the colors were right, but the the placement of the uh bar.

C

What's? What's reverse that's my feedback.

D

Yeah but yeah I was um I was wondering if you could speak to the.

A

um Developer productivity, yes,.

D

So it's hard to quantify at least as I understand when.

A

I quickly go back, uh so I went really quick, so uh there is a lot more to it. So if you go to second tab, it talks about uh developer productivity with and without kubernetes and every stage of the application development Journey. When you, you can do more with less because you have a programmatic infrastructure. This assumes your non-kubernetes infrastructure, probably not even containerized, on microservices. So if you're doing application definition or builds, it just takes more work.

A

Once you have a kubernetes infrastructure, the line items will talk about the number of hours saved and the overall high level message without going to weeds here is that you can ship more bills more releases in less time and that equal equals to making more progress for the business. Therefore, more features out in the market and therefore you know for every expensive level for you. We all the companies pay for work in short supply, you're, actually making them more efficient uh and that's how the value is coming for.

A

You know if I make, if I play a developer, a hundred thousand dollars make them ten percent efficient. That's ten thousand dollars in value right and here make them three times as efficient. That's like 300 000.! uh So that's the basic premise of the calculation.

C

uh Good question, thank you and any other question. I have question up here.

E

Hello I was wondering what the basis of just by taking an application and dropping it into a kubernetes cluster increases efficiency. What's the basis of that.

A

So um the basis of efficiency is, you know, traditionally uh cots commercial off the shelf or traditional virtual machine based applications which are non-containerized, non-cloud native or uh which are applications where black boxes.

A

um So it was hard to Multiplex. Multiple applications on this on the on the same set of pool of cluster nodes uh sensor are full of Hardware, uh so the core premise is that once you go to a clustered system beat anything before it was mesos, Docker, swarm um and I'm talking about kubernetes, because it has become the de facto standard, uh and uh you know it's like dial tone, but once you go to a clustered Cloud native system, because you can Multiplex multiple of apps.

A

Therefore you know you can basically Bend back a little better.

E

Okay, so the basis is from cot applications that expect to own the entire VM yeah to containerized yeah, okay yeah. It could be a problem because most commercial applications are not delivered or don't support running in containers. Yet correct.

A

And that's why here you'll have a business case that this is why you should probably need to do at least that much work yeah.

D

If I do, the work is.

A

It worthwhile and and that then that's a question. We have seen a lot of users and customers like yeah I know it will be useful, but I'm going to have to spend a month containerizing, and this is a way you can kind of preempt that question of like this is clear value and that there's a clear cost now, let's decide uh should we do it or or maybe so for some of them it doesn't make sense, but at least you'll have a baseline to make that decision.

E

C

Any other questions.

A

Cool I think there's a couple of more minutes.

C

A

More questions over there.

B

Hi, my name is Lily I'm with Kube cost I had a question just regarding ongoing validation for kubernetes, so this seems like a really great tool for at least making the first initial business case to Executive stakeholders. But what happens once you have kubernetes in production? You have different application use cases. How would you view fincube as a tool to then further continue to validate ongoing projects over time?

B

A

So the question was that uh you know uh from this lady from Cube cross. uh How do we uh valid an ongoing basis?

A

um The value delivered, so what I uh showed pin Cube as as a toolkit early in the Journey of moving to a cloud native architecture from Discovery to justification to implementation, but once you do that you have to operate and optimize there's an ecosystem of finops, tooling and phenops, and cloud and capacity and availability, optimization tooling, which all of us observably tooling, which you have to deal with, and that was another Focus. So it's not the focus of pen Cube, because there is plenty of work there, plenty of commercial and open source tooling there.

A

What folks should and can leverage you know to kind of continuously operate continuously optimize and keep keep their infrastructure optimized and be able to kind of report back close the loop there but yeah. He was really focusing on early in the life cycle. How do you? How do you set the rules of the game?

A

Another question from lady over there.

C

A

Okay, if there is no other question, you know please take a moment to leave it leave us any feedback, appreciate it join the fincube community and if you have more questions, if you have more feedback, come find me at the su-84 booth happy to talk about fincube, Fin Ops, and what should we do next? Thank you.