The Business Doesn’t Care About Kubernetes as a Kost Center. They Care About This, with Bryan Ross.

Episode 272 · Fri, 04 Aug 2023 08:05:00 +0200 · 57:15 · SRT · archive.org · all Tanzu Talk

well you're back again brian ross uh for maybe your second appearance is that right yes thank you so much yeah a little while i think the last one was uh early this year that sounds about right yeah something like that yeah yeah yeah well we we were we were uh talking with each other through slides and dming and stuff and and you i think you brought up a intriguing question that i i uh i haven't contemplated in a long time it's like nowadays there's a lot of people who interested in a path to production a golden path even a secure software supply chain to do my obligatory joke i don't think anyone wants an unsecure supply chain so whatever but it's always good to emphasize things and uh uh and and i think i think you raised the point that like well i mean haven't we been doing you know uh automated builds since the day if anyone remembers of cruise control and then you had i think it was hudson and then oracle had a trademark little problem and then it became jenkins and then you had uh i think atlassian has some you got this that and the other all the way up to like uh you got github actions that people seem to be incorporating into their things and so like what's up with like uh this golden path path to production talk like are we like uh what what's new here what's what's the consideration you know it's really interesting that um all that is new is old right um as i say we've had a lot of this technology before and i'm out talking to lots of customers about golden path about platform engineering and these terms have got a lot of buzz and a lot of following and the risk with it i think is that we always end up looking at that thing as something new that we should follow in itself um you know kubernetes would be no different here right brand new technology we should definitely deploy that we want to have a kubernetes platform and my job is largely to ask why why are we doing these things what is it we're looking to achieve and so when i think about

golden path i think less about the tools that are in place and a bit more about what the value is we're trying to get and i think that's one of the areas that i'm seeing now and i'm on the cusp of um thinking about a blog article around the sins of platform engineering because i'm hearing a couple of key phrases now that kind of signpost to me that perhaps we're not doing platform engineering perhaps where we're just focusing on too much on the tools and it came from a the idea of a friend of mine so for example you're asking about the the differences i think the the main thing is that developers have been doing cisd for a long time and i think with golden path what we're trying to do is be a bit more abstract about how we approach that so it's not just one way of deploying software because developers did that already with hudson and jenkins and everything that followed it but a golden path for me is the ability to build something that developers want to use in other words you have a cicd pipeline that is flexible enough that many development teams give up their own environments in order to use your central one and that's the critical thing i think there's so many organizations i speak to that have deployed effectively a cicd and they're still disappointed that developers are refusing to use it they know that the developer's own one is more locally optimized does things that perhaps the central one can't do it's tooling they're really comfortable with etc etc you know like what what is your because what's your sense for how many organizations are building out let's say a kubernetes cluster or or a general cloud infrastructure strategy with the intention of it being used to run their own software to be used by developers application developers like i feel like i talk with maybe i'll be a little aggressive or overly whatever maybe 50 of the

people i talk with who are in that situation what they're saying is they've built it and like developers aren't really like as we would say back home cotton into it like they're not really like uh uh attracted to it or wanting to use it like what and and you know there is a selection bias for both of us that we often because we work at a vendor and ultimately we're trying to help of course we're trying to make the dreams come true of the people we talk with but we're also trying to help sell them stuff and so obviously if everything is running perfectly and they don't need any help we don't talk with them sort of but uh so there's a bit of bias there or whatever but like how common do you think that situation is i think it is very common certainly the the experiences i have most organizations they have a a cloud strategy that is normally based around one vendor or more and they have a kubernetes strategy which is weird in itself right like you could you have a kubernetes strategy or should you have a cloud native strategy or an application modernization strategy something that's going to achieve a business outcome um and you know one of the things i love about my role is i don't have to to go sell one of the best things i honestly get to turn up ask what problems they're having and if i don't think i can help i i just saunter off and go go see the next next person right but i think the secret is that many big organizations small organizations ones that are moving quickly ones that are moving slowly all seem to suffer the same problem in that they are i think suffering a little bit of a hangover of the last couple of years of rapid investment and development so i think back over the last three to four years there was a massive boom in spending and around it there was a huge amount of optimism because you know awful things happen in the world but as a result

we ended up doing lots more remote lots more online digital experience all of a sudden became the top priority for everybody so you know it was it was the heyday some in some ways for it but now you know there's the world has changed again financial downturn we have various things happening abroad we've got supply chain problems you name it and i think that we're now beginning to look at some of those technology choices and ask ourselves what did we really get from this um you wrote a great thing on your cotie newsletter um around paas and kubernetes and whether that was a seven-year distraction and i think that's that's kind of what i'm seeing in organizations now is that they're beginning to look at things like that kubernetes strategy or that public cloud vendor strategy and think about how they're going to deliver some business value out of it because they're really getting squeezed for it yeah that's an interesting take on it because it like like it does it it it does kind of fit like one of my one of my fanciful notions of i mean as you were kind of alluding to that uh for whatever reason like we've built out a bunch of infrastructure and and now we need uh we need to worry about what it's being used for and it's it's such a big as as people still say a big lift it's such a complicated big piece of effort to build out a cloud essentially or or you know assemble the way you're going to use uh a cloud or clouds that like you've got to prioritize and maybe worrying about your developer experiences like it kind of makes sense that you would do that after you build the infrastructure right and so like even in the most optimistic case like now is the time after a couple of years of building stuff out that like oh now we need to do developer stuff right which which then also i think you know i don't know it's hard it's hard to know how much of this is just me making up stuff but but it it it would make sense as as you're kind of saying that now there's there's a need to like

show value for it and uh you're not really going to show value by just having like infrastructure like it needs to do something and yeah and the the the the best way to do something with it well not the best way but you know you can either i guess the things you can do with infrastructure tell me what you think there's at least three things one you can do less of it to save money which is always appreciated people like saving money but that can be difficult you know i'm no accountant but the way that accounting works i mean you know you're on three five year amortization plans and the costs are accounted for so like just shutting something down might actually not make any sense because the costs are already sunk i don't really know what i'm talking about but but you you can always decide to do something less with infrastructure to maybe save money or you could like run like other applications off the shelf applications right like in particular maybe you could do like some data processing if you wanted to just use that roar roar raw horsepower to uh do analytics and things like that or you could like make your software your own applications better somehow right and that that that kind of becomes a bit of a to use an overused silly metaphor a bit of a fractal thing where like the way you make your applications better might be to make them cheaper right so you could consolidate applications or make them cheaper or you could actually do your software so that your uh somehow making your business run better which could also be cost savings or you're actually uh i guess you could be retaining your customers better so you could be preventing churn or loss and then so you got negative one reducing costs zero maintaining costs the same or you get one which is like you grow your business somehow by by making your software better so like it seems like the the the logical conclusion is like you need to do software or analytics or do less infrastructure i i guess maybe

seem like the three levers you have to uh show value to show that your stuff is worthwhile i don't know well what's up with my grand tour there did uh is that anything yeah well look i think that's that's kind of what i was saying that the we've we've come from a period of massive growth where immediately the digital experience around the business became one of the most important things whereas a business might have had an online shop before they still saw themselves having physical entities where people would do the most of the business whereas you know the the global pandemic brought us to position where all businesses had to have an online experience and more still that online experience had to be competitive whereas before it just kind of had to exist i think the change in finances though means that we are now looking at that saying well this growth is still definitely great that's how we're going to grow our business that's where we think the next stream of revenue is coming from but we need to balance the the books at the end of the day um so without doubt i think with my customers i'm still seeing that complete dedication to building a digital experience for customers and making sure that's the best it can be yeah i think largely as well like the the kids today uh you know they're all going to want online transactions that's now the norm so i don't think that's going to change dramatically for the next couple of years so more and more online transactions um but i think again we need to make sure we're doing this sensibly um and a lot of this is really basic right i think the it's very similar to the chat we had at the beginning of the year and i'll keep ranting about it until something changes in the world but whatever we do we need to be able to answer the why really easily and quickly so you know i've gone to many customers and said right great you're doing kubernetes why why are we doing this what what part of the business is going to be improved

or even more now i'm going to people that say well we've got a platform engineering team and they're doing great work and they are but when they say they're customer focused and i ask them which languages their developers are using or what their ide is or what are their pains that developers feel every day that the faces go a little blank right so the running joke that i've got just now with a friend of mine is that you know if someone says they're customer focused it likely means they spoke to one developer one time or maybe someone even used to do development a decade ago but we are still very much in the habit in it of building things that we think others will find useful and that's not the way to run a business you need to um you need to go out there and do your market research and build things that are going to be useful demonstratably every day and for me i think that's getting into trenches and saying what really irks you on a tuesday and then fix that problem and that might not be a huge golden path that might actually be one really specific thing like i struggle to get api documentation from other teams right let's go solve that problem right little problems solve them iteratively and then build it into that pipeline if you do that then i think there's more chance developers will find what you're doing is useful and then the more they're going to use it and the more that you can then demonstrate that this thing that you've built is actually worthwhile that it has some value so so let's let's pull at a lot of these notions here and and like so we started off uh with the premise that like what's what's new about a golden path right and which which which you know is a good question and then we have uh so you build up all this kubernetes infrastructure and and and what's it useful for and then we have like the idea of um uh you know a business most organizations to to make it profit and non-profit most organizations like uh are operating with software as a core component if

they didn't have their own software they wouldn't be able to operate as they are today and they probably wouldn't be able to like make their their business or their organization run better you know i mean obviously in each industry there are things apart from software that make their business better like in in in one of your former roles like if you have content that's actually entertaining that's cool right like you could have all the cool software in the world but if your content's not good doesn't matter if if you're in the entertainment industry so forth and so on uh so like there's something something that we're talking about at the the the middle of a lot of this which is like making it matter to the business and the the term of art the jargon for this in our trade is business value which you know is is basically a concise way of saying what you're doing is useful for whatever it is the organization wants right like like it's you know something that is not a business value might be like when we when we talk about the stirring straws for the cafeteria for the tea or coffee should they be green or should they be off white right like anytime you spend doing that unless you're maybe in the fashion industry and one of those colors is core to your identity again there's always exceptions but in general there's no business value to that technology the stirring stick that you use there um and you could argue that like there's not much business value you know you you require a stirring stick otherwise you can't mix the coffee but it doesn't matter what kind of stirring stick you have kind of sort of like a file system like you know you need a file system but it's not really worth debating which one to use unless you're doing something very exotic just like a green stirring stick and i need to cut myself off here before i get lost in this stirring stick thing but like like so so like you know given this like cloud native stack like how how do you think about like a very not detailed but kind of like a a nicely causality linked way of thinking about business value like how do you go from like

a golden path helping developers out to like saying ah this is clear business value right something that we can kind of kind of like track and know not just like trust that it's connected that's a really interesting point um so i think that business value begins with a mindset of serving others because i think fundamentally what you were saying there about um stutters not to get too deep in that but it's it's around what is commodity right so i in my head as you were talking i had kind of worldly maps popping up in my head around what's important to a business and what's not where should you tie spend your time and where should you outsource for example and i think it's maybe similar so computers by themselves don't really do much right they're fairly dull but what we can do with them when we turn them to something that's unique to that business it's a bit more interesting so as people in it the closest we have to that i think is twofold number one is talking to finance to reduce those numbers whatever the spend is let's make that lower if we're spending lots of money in one area maybe we can consolidate and you know vmware made a very successful business out of that consolidation play um i think the second is being able to help grow the business um so the metrics around there is how quickly can we get new ideas to market um and that is that's how quickly can we deploy software but more importantly how quickly can we change software i think that's largely where enterprises are struggling sometimes they can get something new launched in public cloud that has bypassed a lot of their internal processes but then when it comes to scale or when it comes to making a change or making it more secure or auditing it all of a sudden changes have to be made and everything kind of slows to it to a

grind halt yeah so i think for me i think looking at values thinking about what the impact is you have on other teams so developers are our primary stakeholders so how can we improve their day i think that can be largely removing toil from their day um and i think we spend a lot of time thinking about productivity right like getting 10 minutes back per developer multiply that out that's pretty big but i think there's a bigger play as well which is if you can make developers happier in the role so they're spending less time doing this plumbing that they didn't really want to do in the first place well in actual fact you'll end up with happier developers which means they're going to hang around in the business for longer which means you're not going to have to hire new developers as quickly and finding new talent is actually pretty difficult now right um especially at the kind of price point you want so there's an hr part there around you're adding value around engagement for for people at the end of the day developers are human they just want to make an impact if they want to change that stirring stick from green to off-white they want to be able to make a simple change like that overnight in many of the businesses i'm working with that's not possible and i've seen more than one developer friend of mine join a new business and leave within three months because they just couldn't get anything done and i think that uh part of the work we can do with helping developers is really just focusing on how we reduce the time they spend doing non-value work because for a developer their value is when their application does something that it didn't do before it's a new feature it stops crashing that's that's value yeah you know so there's a we'll see if i remember all these things to follow up on in there because because there's a lot of interesting stuff there and well first of all i mean if i were to summarize use a phrase to to to summer phrase eyes uh the business value result or connection you're

talking about there it's like time to market right and so if we work back from the time to market thing which as you're saying like uh the way that often manifests itself in the it space in the software delivery space all the way down to the infrastructure is uh removing wait time and let's say friction from the process right speeding up the time it takes to go from idea to like a user using some software right and um so first of all like the thing that i've struggled with over the years is knowing knowing when an organization actually cares about time to market beyond this kind of like fear driven we don't want to be blockbuster mentality right like you know that that kind of like which which is totally cool if that's the milieu of like the world and everyone kind of like has this like uh this need to do things and i suppose you know i i've i've done a lot of uh discussion here in in tonsuland and webinars with uh your your co-worker darren rice about like this urgency in the financial sector right and to some extent well it does exist there that there's this notion of like at least in well i'll say at least in retail banking i i can see how this urgency would exist in other you know trading and other stuff but like that we need to have software that isn't terrible right that that that people like you and me use for our day-to-day life and uh as things come along like being able to pay with your phone we should do we should deliver that quickly right and like so there's that case and then on the ux the other end like i think an area where time to market the one that i've encountered the most recently is very well understood and i feel like the organization like has actual time to market urgency is like

when you when you hear like militaries talking about their need to use software they're basically like we need to be able to change our software really quickly because it's how we run what we do and what we do is very important it's it's literally mission crit like mission mission critical right uh and so like that urgency and and you've seen this over the past i've seen over the past 10 years that like you know in in various various countries militaries have really like they've been like oh right we're not good at software we don't even have that capability so we need to build it up because we need to have like programmers in the field doing that stuff so like that's kind of the ultimate urgency uh in companies but like but like what i struggle with is like unless i'm talking with someone an organization and they already are just like we want all the time to market possible whatever it is we want to speed up like it's almost like you don't have to talk to them about a path to production or a golden path they're like i want to remove i want to speed this up as much as possible so unless they have that like finding out what their time to market is like ferreting it out and then connecting it down to uh you know you should install kubernetes right like or or you know out of our infrastructure area a very stereotypical cliche thing for a developer to complain about saying why you should get a laptop that's three to four times more expensive than the standard issue laptop right like like kind of convincing then whoever is control of the it budget and like the authorities that like one time to market matters to you and one of the ways you're going to achieve time to market is by doing just as you were saying like removing all the waste from the developer from the path the golden path and then and then like even more foreign of a difficulty is like the thing that you're alluding to is like yeah

like happy people do better work and so like if they want a green stir straw we should get them green stir straws right right like you know kind of like that that kind of stuff so i don't know there's a question in there somewhere but it's sort of like detail more of that like if you're talking to someone and like they don't they're not quite on this time to market thing let alone connecting it to optimizing developers like how does that conversation go how did it when do you see the light turn off where they're like oh right we should spend three thousand dollars on a developer laptop every 12 months not every not spend like 200 not 200 not spend like you know 400 or 400 pounds or 400 euros every three years right or or you know we should buy into this infrastructure that's like better blah blah blah right like it seems like that's a that's an issue so we've mentioned path to production a couple of times um those are without doubt my favorite workshops that we do and i think it's largely because it's not all technology focused and for anyone that's not done one before it's exactly as it sounds we we get as many people from across the business in different parts of the business to talk about how a company goes from having an idea either something brand new or just a simple change they want to make all the way through the approval processes the financial planning through to development hosting testing deployment scale and then then into a steady state um it's usually during those exercises that we discover where productivity gains really lie um sometimes it'll be really obvious where you can see there's huge breakdowns perhaps you have a complicated it process where we've worked out what we want to do we know exactly what we need to to have

but it involves sending tickets to seven eight ten different it teams and you have this ping pong of request tickets going back and forward until eventually a golden nugget drops out the end of the machine um i've seen that far too many times and it's the individual steps aren't usually the problem it's the wait time between those tickets right right right i always joke that the uh uh it's not the arrows i mean it's not that i messed up the joke it's not the boxes that matter it's the arrows that matter the handoffs between between the every these things 100 um so a lot of times again it makes me unpopular perhaps in the sales capacity because it's about looking at the saying well i don't think we need to optimize the it here we need to think about optimizing the path then we can talk about how we're going to optimize the path and i think technology is an enabler it shouldn't be the destination and i think again this is a mistake people run out and build a kubernetes platform on the assumption it's going to make them faster more agile and all of those things but in reality when you look at that path to prod there are months and months of effort involved and even getting to the stage before you get to talk about the platform so how can we get further down that path to try and help out there instead again coming back this idea of what does a developer struggle with day to day i can think of many organizations where they have a brilliant brilliant cit pipeline just ready and waiting to go as soon as you have your container image ready they can have that hosted in different countries various data centers auto scale with monitoring it's fantastic that doesn't help a developer develop most developers when they're beginning to develop have a couple of really basic problems the first one is they write some code they then want to run it to see if it works or how it looks and then they change it and they run it again and this like iterative hamster loop needs to

be really really quick and in terms of getting their feedback if they have to wait two three five ten minutes to see the output of their work it would be a nightmare right it'd be like if you're using a word processor and you type a sentence and then you need to wait 30 seconds until that sentence appears on screen that's the type of pain that some developers are seeing just now because they can't deploy their development code immediately on the cloud infrastructure so they build something locally perhaps on a little machine under their desk we've all been there or perhaps they've got some docker containers running on their local machine that they built themselves and has nothing to do with it then you've got that mismatch between what their development environment looks like versus the real development environments right so how do we help there right and uh you know i won't sell too much but turns out tanzu does have a couple of solutions there and i go out and talk about these five best-kept secrets tanzu because the parts that most people don't really paying attention to but i think they're the most valuable and one of them is it's just a work of art it's a plug-in for intellij and for visual studio code that allows developers to write their code hit a goal button and then see the output running on a kubernetes platform that is centralized they don't have to build the container they don't have to think about the kubernetes infrastructure existing they just have their code in one window and the output in another and they can set break points and test their code and do development things that's how we can make technology a lever for increasing productivity not the fact that there was a kubernetes platform there but the fact you're able to use it without really seeing it but it means they're not having to waste time building their own development environment or fixing it every thursday or friday because they've broken it or tidying it up or trying to work out the mismatches and the differences between their development environment the production one

i think that's that's really getting down to speaking to people about their daily challenges and helping solve them using technology but yeah dot dot technology in itself yeah you know it's making me think you know i always like to reduce things down to quippy little whatever's but like or or painfully not quippy quips but like like a part of what you're describing there is that a lot of the infrastructure that exists exists to run your software in production but it doesn't really exist to be very useful for the developers writing that software and so like you were saying a lot of what happens the the you know an interesting experiment there is to say how different are your development environments than your production environments right and the maybe not so uncommon extreme version is that the developer environments are completely different because the developers have felt the need to build their own infrastructure that is barely matches right they run it on their laptop so they have the the the desk under the desk data centers as you were alluding to right and it's just there's so much friction and maybe also just so little not existing as far as a concern or a thing that happens that the people in charge of the infrastructure just don't really care about what the developers are doing right like about and and they more uh are thinking they say like write your software to run in this kind of environment and give it to us and we'll work out the kinks between the two of us right and so this this creates like the ultimate not the ultimate but it creates a good example of the arrows mattering more than the boxes right to use a a fancy term because you have local optimization at that point like

each of the the development phase can be very quick and they can be moving fast and doing everything great and it can be locally optimized but then once the you hit that arrow and you go to like even qa right like qa is going to have to stand this up and then even when you go to i mean the only reason well not the only reason but a reason that something like staging exists is to make sure it works in in a production environment versus whatever else so you each of these arrows you hop through if you're not running like like more or less the identical infrastructure that the software was developed on it's not cool right like it's it's just more and more friction and i think the mentality is is because like that infrastructure was built to run the applications and not like be not also be usable by the developers which i think is i was talking with with with uh jennifer riggins who writes for the new stack and you know she's been studying platform engineering for a while and we are discussing this morning like what is the difference between devops and platform engineering and she kind of as as you several times she brought up the the the uh the like i this is my rewording of what she was what she was saying but it's sort of like i think in platform engineering you talk with the developers now that again that's my rephrasing of it but like it does seem like that's the jump that that platform engineering people are trying to make and that we in pivotal tonsu land made a long time ago with our platform as a product and with uh pivotal cloud foundry now known as the tonsu application service like that's that was all built around that if you remember i always thought these were a little yeah i thought these were a little goofy at the time but uh our our head of engineering ansi like had these these uh these presentations that were a bunch of emojis and he would he would start to describe developing infrastructure based on understanding the application developer process right and like with they were good presentations

but they were they're also this is before i accepted that emojis were a way that mature adults could uh conduct business but it was it does it does kind of like define a different approach of like we start at the top with the developer and focus on what it takes to get their software and we build the infrastructure out uh according to what's needed there uh which which i think is a good a good turning of the priorities towards not just the developer but towards the software right and and this is the connection i'm always trying to make is like now you are uh you're doing it on behalf of the software and if we think that the software is the way that you implement how the business runs that means you're directly doing it for the business right like you're finally building infrastructure and all of the tools and you have that really is focused on how do we run the business better not just how do we run it better and i think you're absolutely right and it's really interesting when you think about devops and platform engineering because again we everything is news old um devops for me was brilliant we started talking more about helping developers and operators getting closer which was great there was a huge number of new tools that were built out of that time which was fantastic i think the downside though was that it was locally optimized so we were able to get individual teams using cid pipelines and a whole bunch of tooling which was great and we added value in that little silo but the problem is we really struggled to scale that in enterprise because you needed to have qa and developers and operators all in each team so you had massive duplication of people in terms of resource cost but i think as well your duplication of ideas and the struggle was trying to

have the guilds or chapters and various different mechanisms of trying to say but we need to operate in a central way too for me platform engineering is trying to take in the lessons learned from devops and some of the tooling it's about taking um the service orientation from paas and the lessons learned all the way back from cloud foundry and also mixing that with things like platform as a product which is relatively new where we're thinking about not just building out this platform but really going out there and proactively marketing it all of this still points to one big core part and that is as you said we need to talk to the users of our platform and we have to build what is going to make our technology useful we should not and never should build something that we think is going to be useful right you want that validation uh so you have to build that into you have to build validating the service you've delivered and and what you think you should you've got to anyways you need to validate that it's useful no matter what no matter what you thought or what you signed up for six months ago yeah so you know if you're if your goal is to try and get software delivered faster sure you could build a kubernetes platform and make it really quick to deploy that but by speaking developers you might find out their challenges are making that container image in the first place so maybe it's time to focus some effort there um i feel as well actually we've probably ragged on kubernetes quite a lot which for you know a company that mostly sells kubernetes tools and things that's that feels difficult just to balance it out i think the thing here is that talking about usefulness kubernetes in my humble opinion shouldn't be thought of as a platform it should be thought of as a language because that's the value that we've got from kubernetes this idea that a developer can build an image on their workstation and it should run in your private cloud your public clouds your

sovereign cloud or wherever else maybe on the edge right so i think kubernetes as a language has been really useful because that lets us talk about requirements and capabilities separate to the technology that implements them um and so i think kubernetes plays a massive part in some of that modeling that we're doing so developers can come to us with requirements now because we've got this common language we can use to describe what they are and we can build a platform around them as well um there's a lot in there i think for for companies to work with though because again none of this is technology as we know it's not building out large chunks of infrastructure reducing cost necessarily it's about trying to twist some technologies together and bind them in the way that's going to be useful to people much like you if you're building a product somewhere else you take raw materials and you you have a new spin on it and i think that's what businesses need to do they take all these tools in and spin them into a tool that's going to be useful for their business so you know you're so you're you're bringing us back to something i i wanted to come back to as well right so at some point uh when we were talking here you were you're bringing up so you know you have a you have an ide integration for the developers that like make fits into this and makes building containers better and eventually you get down to the infrastructure right and i think i think one of one of the things you're making me think is that as you were saying a lot of what we don't well i think that a lot of what we as a whole community of doing this stuff don't think about very well is that what are you saying that that twisty tie stack right like this notion that we have each of these components in a stack and how how they integrate and fit together basically to do to make your your time to market your ability to deliver software faster to have tight tighter to deliver let's say every week maybe every two

weeks right like just to pick a not so arbitrary period right like i think i think if you can deliver new versions of your software new features every week or every two weeks you're doing pretty good and you'll really be able to like use software as a core way that not only you run your business as it exists but you'll be able to like innovate in your business right and so if you can twisty tie a bunch of stuff together that delivers that then i think that's kind of like the combination of your platform and your golden path and your path to production all of this stuff right and and i feel like one of the things that's often missing is one understanding regardless of the technology what the components of that stack are and how they're supposed to fit together right and this gets to the old like you know a bit of the distraction thing is like i think everyone's like you know myself included early on when i heard all these these big claims about like what kubernetes could do and like things like that like they sort of thought like once you went through the kubernetes the hard way installation you were done right like you would have all this ability and like you would basically just have a heroku finally for the enterprise that that people would want but like we all found out that was not the case and blah blah blah right by design right like which is totally fine but like you're making me think what the conversation i think i would like to have with people is so what is that twisty tie stack right and now let's go box by box and talk about what the best fit is for that now you can also like you know this that's a bottoms up best for breed thing best of breed thing and then that also leads to a discussion of like you know what we do we have a twisty tie stack right everyone has a twisty size stack and then if you can kind of understand what the components are and how they fit together well then i think you can have a more informed conversation about all those

various stacks and the parts of them right whereas now it's like at least to my mind it's very foggy like how uh a build tool integrates with like the way you're doing a kubernetes thing versus integrates with this thing that then like all the way up to like and then i don't know there's a ui framework and like it's it's like a bold claim from people like ourselves and others to be like our platform is the best all the way up to spring and it all integrates together and it's super cool right without like like we almost get trapped in something that i think we did in the in the past era back uh in the in the 2000s where we're basically just like hey here's a platform trust us it's just a it's it's a you can't look inside the box and how it fits together but it's totally cool where it's hard it's like we almost need to have a new conversation where we talk about what these boxes are and i think this is this is an area where like i think there's good parts of kubernetes is it's trying to establish a common as you say language or uh as i would think about it just like a common api and a data standard so that you can kind of like you don't have to have all these custom ways of having the boxes integrate and how they run but i don't know i mean that's like it's i'm always searching for that little rosetta stone i don't know i don't know the metaphor for for what that is and i think kubernetes comes a long way to answering that um like having that common language that common api has enabled so many things and i think this is the value of it it's not kubernetes it's what can we use kubernetes for so we talked about um packaging code that has been a problem since time began it's one of the reasons why heroku did really well in the early days um it's why cloud foundry was great as well you had your php app your pearl app your java app cf push and the platform

would work out what kind of runtime you need based on the code language you're using and host that for you which was pretty phenomenal and it's taken us a little while almost to get back to that point with kubernetes with cloud native build packs we can now do that too so that's great now we can have a developers don't need to build their own java environment for their container we can have a centralized one and we can put all of our good corporate tools in there and that's fantastic and we're making the developers like easier in the process we're not asking to do anything else or anything more in fact we're asking to do less so that's kind of cool i think as well when containers first got on the scene like i'm old enough to remember jails and solaris and all that kind of stuff but i think with containers what i saw being a guy who was running a big virtualization platform at the time was introspection being the critical part and i immediately jumped forward five years in my thinking which was if i can see what's running on my platform that i'm hosting i could deliver a whole bunch of new value back to the business right i wasn't thinking about smaller or faster i was thinking about i could make containers way more secure than anything else because i can answer challenges that i've had in the past like we've got a big security vulnerability that's come out how vulnerable are we you're running a platform there with 50 000 vms which of those 50 000 vms is vulnerable to this attack and as a platform owner i mostly had to shrug my shoulders because i didn't have visibility into the what's running in each virtual machine with containers and kubernetes we have that i think as well we can do so much with that as well so we've got things like um being able to look at the network connections between containers looking at the api that is set up between one application and the other and if we can detect apis with like open api then we can build dashboards if we can build dashboards

of apis that are already there and running well we could do some cool stuff there right that would be things like helping developers understand if someone in a different team has already written a piece of code to solve that problem you're having and this goes back to the problems we're having 20 years ago it's developing code is something that's definitely doable but how do you know whether you should develop this piece of code or not maybe someone's done it already in fact your business probably has three or four ways of doing that same thing and it's just because steve didn't know that sarah's already written a better version no i can think back to it's always steve from right um i can think back to a business where you know there were brand new products launched and one of the core components of that new product called back to an ancient deprecated system and it's just because one of the developers that's how they did it on some other project they worked on so they called this old api rather than new version and there was lots of scrabbling around to get that function swapped out but that was a you know that's a brand new product launched to millions of customers and we'd inadvertently pinned it to deprecated infrastructure that wasn't reliable wasn't easy to change and all of those other problems so you know i think with kubernetes we've got this ability to see more what we're doing we've got this language or this api that we can use to describe things better and make it run elsewhere i think my big challenge to the business is well you've got all these tools now what are you going to do with them that helps someone else and then what right you to build value over that yeah and i and i think i think you're arriving at an interesting point there as as we try to quest out let's say what is the business value of like a cloud native stack which nowadays basically means kubernetes right and i think you know it's it's it's easy for me and and you know maybe all of us in

in in this kind of like uh frenzied this stuff is important thought leadership area to get caught up in which is like you know the complexity of things and the difficulty of it and then and then also on the other on on the other hand kind of like keynotey stories about how like things are awesome right like we installed this we put this in place and now we can deliver multiple times a week or like now now we can do that and i think something in the middle that you were getting to is is well and then there's also the the the fear-based preventing mistakes and and and cost controls and like you know stuff like that like like you know we need sovereign cloud right like there's there's these kind of like i forget when i was talking with with with uh darren about this what we categorize these as but they're just sort of like things you have to do that you would rather not do like regulations and and you know there's that kind of stuff and security like of course you don't want the bad things to happen but if the bad things weren't happening you wouldn't do them because they were fun right like um but i think there's something else that like is even more important is to when it comes to business value is to think about so when we introduce this new way of working or this new technology when it comes to software what we need to focus on is what new capabilities we're going to gain that make things better right so so you started off it's still still a very technical thing well very relatively technical thing of like so i when i was doing uh you know vms and and early containers and stuff i gained this new capability of being able to know what's running of kind of being able to diagnose and introspect things so that was a valuable new capability because the ultimate thing you want your it to do is to run and not fail that that's always great and so like if that is a problem then it is of extreme business value just to keep the stuff up and running right to to just be alive so the ability to diagnose things faster and know what's going on is

very much so of business value so it seems like uh the question with building out whether you're you're spending the time to build out a golden path or uh build up some like multi-cloud strategy or build up a kubernetes cluster with the platform on top like it's almost like what you want to start with is this kind of imagining scenario of like all right so once we get this stuff in place like what new capabilities are we going to have and then also as you were saying like is anyone going to use them right like so like it's a two-part thing of like what new capability well there's three parts what new capabilities are we going to get right how will we get people to actually use them right like because if they don't use them then it doesn't matter and then three you've got to do this almost and this is kind of what a business case kind of sort of is but it's hard to reduce it to numbers it's like and is the are the cost the effort the time and the risk that it takes to migrate over to this new thing going to be so much better than doing nothing right like because chances are if you're considering this and the type of large organizations we talk with like you already have a lot of everything like you're already functioning and running and so you've really got to go get over this hurdle of like or we could do nothing and the business still operates right so like so like the capabilities we gain have to be like way good like way better than what we're currently doing or you need that big existential thing of like well what what we're currently doing just does not work so anything would be better that that that we would be doing so i've had a number of comments we mentioned um cloud foundry and tanzo application service um earlier that's been really interesting for me talking to big enterprises and i i speak to them and we say well we've got cloud foundry running with through we had pcf originally and now we've got taz in there it's been running but it's it's a decade old so we want to replace that with kubernetes

and i have that unfortunate job of asking the why which was what value are you going to get by moving to kubernetes and if you've already adopted a paas platform well an actual fact that the improvements can potentially be marginal because you maybe already have a really good centralized platform that enables developers to push code in a centralized way to secure images and that you can scale and that you can monitor like in a in a sense it is a container platform in its own right you know um and so as you say is it is it worth making the leap what is it that you're going to get from a business perspective that is going to make all that effort worthwhile and as it people we always have this assumption that just because something is old the new version must be better and that's not always going to be the case and i think it's really important that we we pause and really think about the impact what we're going to do the flip side by the way is also true i've working with some businesses just now doing very cool trendy edge stuff in reality what it is it's things like they've got factories or they have um in case of retail they'll have stores and these places generally don't have a data center themselves they've got computers stacked up in a corner somewhere and the broadband connection maybe isn't great well that's one of the areas where actually deploying kubernetes could really help much like the virtualization space it gives you the ability to run more than one computer on a physical piece of tin kubernetes gives you an ability to operate those remote sites similarly to how you do on your your on-prem stuff or your public cloud stuff um and it's really interesting because some of that's really low weight you like you don't want a big huge heavy environment you want something relatively lightweight but that enables you then if you've got that compute power that is flexible well now you could deploy new applications into your factories or stores

and what could you do with that that could be things like computer vision it could be things like 5g in factories for tracking where things are you know the the end the world is endless there so i think it's about working out where how the technology is going to be made to be useful you're either going to make lives better for someone or there's going to be a new capability that you are you're now able to do that you weren't able to do before yeah well i think that's great that's a good place to arrive at is is is start with a conversation about what new capabilities am i going to is the organization going to gain and kind of like navigate your way through that and and you know that can go all sorts of directions but i think it's it's a starting spot that that uh i don't see a lot of people starting with that's a great sentence right there well uh thanks again for being on it's uh it's great if people want to uh uh well you know i'll recommend something you you're you do a pretty good job of of of one of the the good tricks of people whenever you show up in a video somewhere you download it and upload it to your own youtube channel so if you go to if you go to your youtube channel which you could also do with a playlist but i download all my videos and store them in case they disappear one day but uh you're you're uh what are you you're you're brian ross uk in youtube that's right and and uh you can see all of the uh the great appearances that you've had uh including you know you have your uh your shopping cart discussion from explore uh barcelona last year and a few other interviews here there that i think are uh worth checking out well thanks for being on i think i think you always bring a uh a good like what's the point of this like like angle and then also what's the point of this right now uh to it so if uh if you want to see that uh that youtube channel and other things of brian's uh if you go to tonsutalk.com you can find the show notes and then also uh you know if you're watching this in youtube or uh or linkedin uh i'll put some links there as well so with that

we'll see everyone next time bye bye

Tanzu Talk · 🍕 Get more pizza.