Change Your Mindset with Platform-as-Product, with Stark & Wayne’s Dr. Nic

Episode 115 · 47:22 · SRT · archive.org · all Tanzu Talk

so the question of the day is why is platform as a product or maybe platform as product and you can drop the article why is that important so um to introduce our guest dr nick uh stark and wayne actually there's a big practice around uh helping folks get um pcf installed up and running but as part of that journey uh the kind of change in the mindset towards hey this isn't just like another piece of infrastructure software that you install and run it's it's sort of a new way of living if you will um correct me if i'm wrong that's kind of become a big part of your practices as well so stark and wayne is a consultancy around uh pivotal cloud foundry we came into existence at the same time pivotal came into existence and the mission was was aligned such that for the last five six years you know we've sort of gone to market together helping our customers what we all know is that running you know your platform whether it's the pivotal applications service um there's or whether it's the pks whatever the whole aim is to rechange their internal customer mindset you know to they're not just running microsoft active directory and you use it right you're there to run an entire experience for your developers your production team you know you're there to be the the first and last place that they come looking for an experience now most of that experience is part of like the cli or the dashboard or you know now the other part might be asking for support but more and more of you know what we've tried to do over the last half decade has been to make operations make life cycle make everything

about their production apps being successful you know productized experience okay tell us a little bit about the before state when you go in to work with a customer um what are they what are they coming from day one the customer is never thinking about awesome we're gonna have a platform it's gonna be a product this is the best thing you know they've got some terrible pain that they've lived through it for the last two decades the complaints around applications not shipping you know they'll spend weeks and tens of weeks on an application and they still haven't gotten around to putting it on infrastructure or whether even if they're modern day where they're perhaps using modern legacy things like you know going straight to amazon or you know or straight to azure on low-level bear machines they've still got that all that work to do and it's they look at it as work they're not looking at as an experience that they can rely on that their organization is providing them you know it's a battle it's a battle between the engineering team who think they just get to write application code and move on and and some IT team who think they're going to receive this football over the fence they have to go and put in production it's it's it's it's it's trying to say that look there's a line this platform is a line and you the engineering organization and now not just responsible running code but you get to enjoy the responsibility for your applications being successful in production okay yeah that was interesting how you you said enjoy the responsibility it's yeah look some people um look my history before four stock and wayne as i wrote would be on rails web applications and uh i like to joke now that when you start a new rails application and perhaps the same joke if you said for the spring applications you know spring boot bootstrap you you know you get all your scaffold i think it should come with a

big yellow warning at the end that says warning you are about to create a production application do you wish to proceed like how fast we can make new applications can sometimes confuse that you've now got a decade's worth of love and attention to give this thing as it is a i'll be a computerized living breathing thing um now i think now my honest belief is you run your you know you put your application code on a platform most of the the sweat and tears of keeping that you know it does go away um in part because of that experience the cf push you know from a cloud foundry platform perspective you know it's a cf push experience um but also it's the constraints if you can guide the developers and say here are some good constraints we're just going to use my sequel we're just going to use rabbit mq we're not going to pick the coolest thing you've just found out about on you know youtube this morning or twitter this morning we're going to stick to this because if we if we stick to these these well-known well-operated constraints then you will have a productized platform that you can you know live with you won't have to keep asking for favors you won't have to keep finding your favorite it person to set a machine up to turn on mongo or you know whatever um and you can have that that clear line in the sand where it's you sit above this and you will have a good experience if you're below it then you'll you know um you're asking for bespoke things that are untested and they're going to lead off and you now have to look after a bespoke system your code is not the only thing that is yours yeah and this is i think so you know the topic is platform as a product it's it's by if if a developer can just accept that and not ask but why can't i just do bare metal why can't i just do bare amazon why can't i do bare kubernetes if they can accept that they

will have a pleasant life for the next three decades as an engineer the moment they want to take that step down and say but i i want to run you know uh mongo i want to run seven different things or i you know we need this um they better understand that they they're no longer just about the code they've got an entire breathing system their code plus their other code plus all this the versions plus the the every version of kubernetes or every version of you know the underlying packages everything that changes is a change to their application yeah either they accept responsibility for that or they they sit on a beautiful line on a platform i feel like there's an analogy here um forgive me i'm prone to these but i feel like there's a little bit of an analogy with you know trying to get these developers to kind think out you know for the next decade of this code that they're about to push into production think about the next decade three decades of their life and how much time do they want to be spending uh patching and updating and waiting for uh tickets to go back and forth versus you know if you accept that you know you know you could if you live within a few constraints you can live with it the analogy i'm thinking of here is like saving for retirement right absolutely that's a good one so you know if you could get these young people right with their first job be like yeah i know you want to go spend that money on you know uh fancy cars or you know trendy clothes or just drinking money i don't know but if you just set aside five percent of your paycheck like you're going to be in a way better position in 30 years so so so you think that if if some people get it but if we can teach people about compound interest and saving then then we can teach people about platforms i hope not i say that because so few people are good at it i know i know i hope that's not the metaphor i hope that's where the analogy breaks

down and it's easier but that was gonna be my question how do you get how do you convince developers that hey this is the best for you do you have to let them sometimes make mistakes and then this is the best approach yeah i so firstly let's for anyone that that is asking questions should i be saving for retirement oh my god you should be saving for retirement right this is we're now going to cut over to the personal finance section of this podcast that was a little psa from us to you start saving you haven't already vanguard s p 500 fund please please do that yeah um now back to you know the how do you convince the developers right how do you get them you know do you have to educate them on compound interest or is there some other way i guess it's the technical debt right yeah this is the uh and how that compounds it's the reason that we use the for debt and the phrase technical debt is the interest and the compounding interest um that the organization you know your friends your peers the people that perhaps you haven't met yet are going to be paying um and this is where it becomes the organization's partly responsibility to to make it easier to use the platform than not to because then their staff will have that good behavior without it being a major choice and which is like uh perhaps we go back to the personal savings um having money automatically taken out into your 401k or in australia we call it superannuation you make it easier to save than not to right so you're not going to have that battle on a on a monthly basis totally yeah i know it's like where hey when you sign up for direct deposit right right and what's the what are some of the organizational equivalents that you've seen work to do that because this this question just came up for me the other day of you know i need to make getting things on the platform the path of least resistance right was absolutely and so what have you seen work for that what works so i mean a lot a lot of things like cloud foundry make once you've got cloud foundry and you've used it you know it's very hard to

argue that you go somewhere else today even like i i was deploying things to pws the pivotal web services you know you'll run dot pivotal.io yeah we're stuck and wayne many of our little apps public apps our blog our website all going on to that and so you know um we're trying to upgrade an app there and at no point do you ever think well i'll just go and boot up a kubernetes cluster that's probably more fun you know if you're thinking that you've got too many others you know you don't have enough work to do um but when it comes to an organization making them aware that this experience exists that it's pleasant that and learning uh there are things you can do you can i mean you can market internally so often a platform as a product needs an owner and so where it's successful is where there is at least one human being who owns the emotional responsibility for this being successful they are going to sell this thing now they may have a team they may get people to all the things you would do if you were an external business you would want referrals you'd want to get teams that can show success and get testimonials and get them to talk to other teams if you're lucky you might get top down you know you know we've certainly had customers where it's been all top down right cgo comes in and says it's all going to be on um pivotal cloud foundry and all of you know all the teams throw away their previous deployment patterns right um not too begrudgingly often uh and move to to pdf um but if you are an owner and you've still got a sort of battle for acceptance then then think about it as an owner think about it as i'm going to do sales and marketing i've got to you know you may not have to actually sell anything but you may need to get them to go through trials hack days um demonstrate the engineers and their leaders that it's not just a pleasant day one experience but your applications are going to have a pleasant you know experience the next decade or so yeah is there a way to identify that product owner

in your experience have you seen kind of anything that helps surface like ah that person they get it and they have kind of i think of it as almost like the entrepreneurial spirit to sort of see that person you kind of hope that person i would hope that person was discovered in selecting the need for a platform i mean if if if if if a platform pcf is going to be installed without an owner then i i it's just going to be a piece of infrastructure like you know i mean we've perhaps all heard of stories where it goes off and buys a big subscription or something uh it doesn't get installed doesn't get run it's called shelfware and shelfware and so uh yeah how do you find that person that that would be i i think a really important part of of the onboarding of a platform is this if if the people who are the most excited are not also going to take responsibility for its ongoing success either they need to find that person they are all going to look silly and this is any time you evangelize and want to bring something into your organization what are you going to do to make sure it has long-term success that everyone five years from now is talking positively about you and your life choices um and and and that's going to mean that you as the evangelist bringing this this in uh either need to be that person or you need to be finding that person who is going to you know wants to have that an important career opportunity to to make something successful in their organization and then you perhaps as the evangelist need to take the job of or as a senior you know cto or senior person to to to push everything out of the way and to you know get people to pay attention to you um yeah i've heard of customers actually branding pcf internally giving it a name which builds good product loyalty in terms of the product team but it's also makes it easier to market that and you can code it anything you want and as you can theme uh again pcf or not but definitely the pcf console can all be themed the login page can be

themed you can integrate it in with your own active directory so everyone's automatically got accounts um and uh yeah whatever it takes for you and the team to to have realized that this is a huge win a huge win and look it's 2019 um there is so much material now for those for that product owner to to take and to use and to turn around um but yeah i mean that's that's certainly something pivotal perhaps could help is help product owners these these people that own platform as a product to meet each other and find out um and i guess SpringOne is an example of where they can find other product owners the cf summits um uh pivotal days these are hopefully places that they can find other product owners and perhaps swap ideas on how to be successful share notes because they're completely independent right you're just trying to help your organization have success and swap ideas of what works yeah it's a bit lonely right because it's not like hey developers where you know there's maybe not an army of us but there's there's going to be teams and teams um but when you get down to the platform operators it's already a small team and then everyone was already doing something there who's really the owner is like probably the only owner at the company hence the name owner um and so they're it's a it's a bit of a lonely path that they've got to kind of afford but what about what about things that you've seen that are kind of like seem to be missteps or um you know uh just just potholes that people can hit well let's start with the first one the first one is not having at least one product owner right let's you know it's going nowhere um who's going to be the one that that keeps getting groups to come in and and either to put new projects on it who's going to be the one that that puts uh that does the the app transformation because if applications haven't been on on on on on a platform uh and again we'll take pcf as an example there are those constraints

um not all apps are going to just drop in so it's not just an it job of saying all right let's move things off our v sphere vms and just drop them into this right there might be one or two changes and uh and you might want to set up ci i mean there's if i can take a new path on this topic right it's not just there is owning and operating an application the platform is a is an important starting point even with the platform there are still things that change that break your application or are going to leave your application behind the biggest one at the moment is uh we take pcf for example is we're dropping support for uh the trustee ubuntu trustee based applications um it's internally referred to as cf linux fs2 which who knows where that name came from let's not overthink that but the new one's called fs3 so that's exciting um but at a linux ubuntu level it was a bunty ubuntu trustee versus a bunty zennial and uh and we're not going to support both for very long and so um most pcf environments will have had a short window where applications could have been deployed on both but they need to redeploy them to support right so inside these accounts this product owner then has to be worried about this deliver this message and help shepherd shepherd them and and some projects some applications do lose their primary application team they kind of go into maintenance mode yeah that will happen and that is not just to be ignored because someone's going to need to keep a track of and it's not just that we went from trustee to to zennial that might not change anything because the build packs keep upgrading but certainly one of the projects that we work on uh we use ghost which is a node.js blogging thing in order to make it work we needed to get the mysql shell cli in there to do a little bit bumping of some things on startup well that didn't just nicely upgrade from trustee so there are these very sometimes little edge cases and if it's not at the environment level it might just be

that you're using spring and all the different libraries that you use and they're going you can't just not keep upgrading them so even though the platform has its level where you know you don't need to worry about the upgrading of the life cycle of platform or kubernetes or you know of v-sphere or the machines underneath or wherever the entire stack is whilst you don't have to worry about that your application is still filled with these things that need upgrading so this is where an application needs ci needs to have this way for the product owner to be able to you know find out where the versions of applications keep working and um and so you know in our pivotal universe we have concourse but people like jenkins whatever it is that part becomes as important because it allows you to reproduce deployment for a new version check that it works um and perhaps then escalate to hey we didn't you know this this this application that doesn't have a development team anymore ah we need one temporarily please in the next three months yeah i mean and so if you have tests and other parts of the path to production you know developed and automated someone else could come in absolutely and at least figure out listen we we're gonna need to make a few changes because something's going out of support yeah if you care about and there's a pipeline if you care about upgrades and you care about security and you can't care about security and not care about upgrades um you just follow the logic through these applications never get to be passive so even though you think your dev team's finished its work and it's moved off to make other apps the application goes on forever you can't just orphan these children um and now it doesn't mean you need full-time teams but you you as a you as the application owner and hence also the platform owner need to go about planning for you know cf push might not be the

only thing they should be doing it's all the life cycle around the app for the next decade to make sure it's always latest version is being tested of everything otherwise it just falls further and further behind uh and then one day you just cannot upgrade it anymore yeah and you live with that security so like and that's if you care about security and who cares about that maybe you don't yeah it's just it's it's a nuanced the platform does a lot but the application owners have still some responsibility for next decade just the platform makes it a lot easier and a lot more obvious so what's interesting is like it seems like you're drawing the circle around the scope for the platform team tell me if i'm wrong the circle is like okay you've got you've got cloud foundry maybe at sort of the the heart of it right it's kind of a a big chunk of the um automation kind of sitting there on top of all the underlying infrastructure but as a platform owner your your circle is actually much bigger than that and it's including the the path to getting on right it's not going to be a cf push it's probably going to be through a cic oh yeah look if um if if you as the owner are going to own the platform and all the applications sit on it for the next two decades if not use someone else just think of that right and hopefully hopefully come back quickly to to your question then you will have first one app and then many apps who start to look suspiciously like they're falling behind and the people that worked on them disappear and uh and and and and if you go about trying to fix that later on it's going to be on your budget uh because you've got apps that you know need touching fixing cleaning up whatever um you know i think the benefit of being the platform owner and is that your product has a very clear line of a circle or box around it and i think if you know that that box in advance you can be clear hopefully

hopefully i'm being clear you can be clear about the responsibilities of your customers these responsibilities don't disappear if you don't use a platform i think what happens is your life is so convoluted and complex you just never have clear thoughts about these topics wait okay say that again so it's not that these lines don't exist without a platform it's just that nobody's paid attention to no one's taking the trouble to draw them anyway no it when everything else is so busy you know um there are certain problem spaces that you just hope never turn up okay um re-ability to build machines we go back a few years we were all getting very excited with chef and bosch and things like that our ability to rebuild machines that allowed us to not fear about machines disappearing whereas before people used to celebrate this machine has been up for six years right oh my lord yeah no idea how to rebuild this in the event of of a disaster or the need to rebuild this machine yeah all that because you've celebrated the wrong thing you've lost the knowledge and the tooling of how to rebuild it and so uh the people who got that job had so much work to do rebuilding environments uh they did it manually and hence we all got very excited by chef and then got excited by bosch docker and and kubernetes because it's each time that we got more automation we realized we needed to toughen our muscles and toughen our tools and toughen our approach to our profession so that could be reproducible and it could be upgradable and it could be continuously secure and it's one thing i was not when we started you know with cloud foundry and pivotal five six years ago the security story wasn't the primary story but it very quickly became you know almost like the thing we were awesomely good at you know our ability to update stem cells which are the base images you know the what an amazon machine for example ami our ability to update build packs our ability to update the the the stacks you know the the of where the containers run um it became

this calling card the ability to run pks which is you know uh on-demand kubernetes when people start to realize that one kubernetes was not the right model you can't just share one kubernetes across a huge organization you need to have lots of them you better have a tool for you know looking after all of those right um yes the operational yeah so i think this era has been here where uh those of us who get to stand on sufficient amount of orchestration can see that it's allowed us to start solving on a consistent basis you know the the security issues which the news will tell us um any newspaper that hasn't got donald trump on the front probably has some it hack uh from your favorite brand and so you know this this story is sort of becoming as as public and a major newspaper issue is as it is our space to solve yeah and and not to get too political but sometimes donald trump and the hacking go hand in hand oh look at that that's a whole nother story just spelled it right out there jeff yeah sorry i couldn't resist so i wanted to ask quickly transitioning i wanted to ask about this concept of so if you're a product owner that means you have customers your customers being the developers and we talk a lot about here listening to your customers and incorporating their feedback into your product development does that apply to what we're talking about here treating pcf or a platform as a product how do you think about incorporating the developer feedback what they want yeah that's a really good question um because you know if you're the product owner of something you didn't necessarily build you know you you know if you sort of saying you're a french if you're a mcdonald's franchise owner franchisee owner like and you get feedback it's like there's only so many things i can change but thanks for the feedback right um that is a really interesting you know paradigm of where you participate in that um and fortunately i mean not that you can necessarily change things immediately because it's a pre-existing environment there are certain things

you can do on the outside like you can help with the ci you can give them feedback of of how to go about doing ci which isn't necessarily part of the platform but hopefully it's either been a problem solved by teams other teams at your organization or other teams outside and again i i would have to you know perhaps we should put together a big long list of of SpringOne talks about you know applications and ci and looking after them and keeping them upgraded to help feed that information back about you know um but hopefully yeah i mean if not that topic then hopefully any topic they get there is information where that question has been asked to another product owner that product owner has answered it gone to a conference and talked about it hopefully we have a good background of you know this is the you know the awesome part about the you know cloud foundry platform experiences it's not shiny and new like you're you're running something that others have been running for half a decade you know successfully so um so it might be that you know you're going to bring them you know information from somewhere else it might be that you're going to take that um help them you know perhaps build a layer outside our new service broker like say they want a new service and you go okay we can do that we've already got oracle on the inside um it doesn't look like we're ever going to get away from it so let's bring a service broker in so that you know you can continue to use it uh so that's for anyone that's know a service broker it's part of the plug and play nature of pcf it allows you to bring in external pre-existing services make them nicely available in a way that they've never been nicely available before and this is something just because i think a lot of folks often think about um at least in in the the pivotal universe of cloud foundry they'll think of the service brokers that we have in our uh service pcf services marketplace that are available on say the pivotal network for download and those are the ones that are you know we've worked with a bunch of third-party um software vendors you know monitoring tools security tools databases what have you there they've built it and so it's a supported service broker

but but you're kind of suggesting like part of what you are part of what you're doing as a platform owner is actually writing your own service broker it's not integrating i'm going to download it like those are great environment there might be something that is someone one of my customers needs it might be a networking change you know it might not again it might not be yours to change but yours to take responsibility for it make sure your environment has access to this uh special subsystem inside your company and there are more and more ways in which this can be done um not necessarily can you perhaps easily do it it might take some thinking through but like isolating a portion of your pcf environment uh which only then has access to perhaps a sub you know a certain virtual network um it might be yeah integrate it might be installing a new tile so it might be they've found out that hey there's a hundred tiles uh different you know tiles that they can download and say can you please install this one we don't like please like we'd like to be able to test it yeah um it might be um you know asking for help about how this the java build pack works it might even just be asking how java works and you have to explain there's a java build pack let me explain how this works right well doesn't this relate to the beginning of our conversation when you talked about you know providing a consistent environment um you know this is what we're going to use we're going to use you know my sql we're not going to use all these other databases but then you have all these options that something like pcf gives you so how do you kind of balance the we want to keep this clean consistent environment with we want to be responsive to our customers developers and their needs and requests it sounds it sounds like it's a little bit of a balancing act it is and as the product owner you might have a team that runs the platform for you you might have someone like stark and wayne who do it as a managed service uh sort of rent a rent a crowd so we will run the platform for you and we'll you know give you the you know that you say hey can we try this tile app we'll install it for you we'll set that and so

we don't you know we we do what we you know we preach what we uh practice what we preach so we'll set up a pipeline on how this tile gets installed so that new versions get upgraded so that we can look after it setting up backups um this is where i you know earlier on with our finance uh little segue this is where i go on my you know disaster recovery segue it's like if your database tile is installed and you don't know how to do disaster recovery it's just a cache like you as the platform of owner are responsible everyone else assumes you're responsible for making sure all their data is restorable in the event of disaster high availability is nice too but my goodness you better not with a straight face say you've got a platform as a product unless you know how to do disaster recovery yeah so you know this is not necessarily where you go out and sort of say hey you know okay san francisco i'm currently in san francisco uh with mine and uh yeah coming to you live coming to you live yeah obviously it's actually but one of the things that san francisco fire department loves to do is to drive around really noisily just to let you know they're out and about being a fire department um and uh so similarly you you would need to uh and you just trust that the fire department when they're not doing that is practicing putting out fires or picking up sick people i think they do a lot of that too yeah um similarly you as a product owner need to toot your own horn and your product platform's horn about what it can do make sure people know because it's otherwise invisible infrastructure but you better know how to run your platform uh when things go bad yeah so make sure any tile you do want and this is why i say let's constrict ourselves to a subset don't you can expand it but make sure you know how to do disaster recovery make sure you know how you're gonna upgrade it um yeah this is what we focus on as stark and wayne is running platform running the tiles and so um it is kind of fun to sort of yell about it as if it's not my problem but this is you know what we care about a lot well and there's like to to jeff's point

where it's like you're you're providing constraints but then you know you do have this element of optionality i think i think josh mckenty actually had a uh a fun story about this about um picking from the middle of the book in terms of the gear size right so not having too fine not having too wide so picking there's like this nobody knows the number but there's you want to provide some options but you want to constrain them so you need some guardrails but you you can't you can't literally take someone down to like thou shalt use my sql and that is the only data store that you will use like that's going to be a little bit crazy but the point is i'm going to provide you with a short list right and you're going to be able to do everything you need to do within this range of options right at least now it's a conversation but we're not talking about literally everything in the no universe that you can go out and try and do and then someone has to support it so this is perhaps one of the benefits is is that conversation is forced and state databases state is super important to have this conversation if if you just give your end users kubernetes we just give them open stack we just give them amazon accounts and let them have fun they will install databases all by themselves and they will think their job is done they go look my app works i'm out of here have a good weekend uh and they will not have set up backups they will not set up disaster recovery platforms they will not have a run book of what to do in the event of something terrible they will not know how to go back to an old version of your data they will have just satisfied themselves that their app works so part of part of the the benefit of a platform to the entire organization is a nice experience to the end user whilst you're still capturing responsibility and so the ownership of each of those data tiles might be separated into different teams it might be all the product owners but it is still a known responsibility and so stateless apps have fun on the platform stateful things oh my god

please work with your product owner to make sure they're well looked after um it's okay to want to use kafka that sounds like a great idea let's do it together on behalf of your product owner let's do it together so that you have a good capture experience where we have already organized support from uh confluent in this case who probably you know maintain kafka or any of nines or someone else who's got a tile for you know kafka uh you know it's a conversation and yes product owners should should accept those conversations otherwise he will get shadow it and he will get people just going off and doing it themselves yeah okay so so clearly having a product owner is important i think we just spent like 15 minutes on that are there one other no that's i i well i is there are there any other assuming you already have that then what else yeah what else what else can go wrong what else can go wrong right if or even if you're just talking on yeah and i've probably spent a bunch of time talking about this so if you want things to go wrong like you're down the road your abandoned apps will have problems and but if you don't know that then you won't think about it in advance so i i do apologize perhaps for spending time so okay you've got your new customer your new engineering team they have been told there's this platform it's got its own cool name um what can go wrong is they might have had expectations about how deployment works from something else they read right they and this is something new to learn now it's pretty easy to learn but but if you don't actually let them learn if you just sort of throw it at them then then they're going to perhaps react in a negative way and all of a sudden they talk and they get angry and and engineers like to escalate their problems and so we both don't treat them well but also we over over nurture them um i love engineers i'm an engineer i just i'm concerned about my

profession and um so you know whether it's you know a hackathon is an idea just nurturing them through until they have success and then hoping that you know help them teach everyone else um but you the product owner may may need to capture a lot of information about how their java app will be deployed or how to quickly get them that answer so in that early period often we do uh application dojos so you know they really want to take advantage of those initial application dojos they do with with you know the pivots um because you're going to learn about how to write apps the pivotal way but also very importantly about how to work with cloud foundry both both as a human being cf pushing but also starting to set up ci i personally so whose responsibility is it to set up ci by the way this is my personal yeah is it the app yeah it's not even my personal belief it is mathematically proven if you can care about security updates of your application for the rest of its life then you need a ci and so but that doesn't answer the question of who who is setting it up because yeah just to some extent you could say well the platform team seems to be accepting a lot of responsibility you know i i don't think it's certainly it's not the pdf platform the police platform the apple doesn't take doesn't say here is ci so uh and it is kind of like saying hey everyone has to use java that's kind of a step too far perhaps in the relationship and certainly not something that pcf as a platform enforces um some companies do love their favorite ci there are a lot of them and i think to sort of force an organization to use one versus the other might be a step too far but here's a nuance to ci i i like to call ci a little robot a stack away and i think we ended up like we had like 90 something little pipelines that did various things often some of them were benign they would take a package they take a a new github release from a project that wasn't ours would build it and turn it into an asset so we have a um a debian repository uh at apt.starker1.com and it's full of

of things that aren't even ours it's a lot of our stuff but a lot of other people's stuff because they just weren't publishing debian packages so we have a lot of pipelines that do this um and uh so i i know what it's like to look after pipelines but i know very clearly that these pipelines run my business so they're not benign they are important business logic and so when you choose your ci system please know that if if if your important business logic is not going into a git commit and it's not being tracked and you perhaps go even further go even further do you know how to test your pipeline in a in a safe place that is non-trivial ci is often like shell scripts it's very uh destructive like it's doing something and it's very hard to set up a safe way to test that um so you know there is a lot there is a lot you know just like the the idea we were talking about earlier about the importance of always upgrading because you always want to be safe and secure and just security as a first-class thought is easier now that you have platforms i hope in my heart that people start taking ci seriously and it's not just like as a it's like code and if it's like code how do you test it how do you update it how do you because the moment you touch ci it changes everything and maybe it does something disastrous and you go well that was unfortunate no it's not unfortunate you did this to your organization but so i i would love it if there was one ci that we could refer and knew how to do unit testing of ci but i i don't know of any of them it's just a hard problem yeah okay well it is the application owner's responsibility okay thank you i was going to say it's like i agree it is someone's responsibility and it's the application okay application you should come back to them 10 years from now and say you know you made these life choices and they went terribly wrong regardless that's a whole other topic how do you get somebody to care about 10 years in the future when chances are they're not going to be there anymore

right and this is where our platform i think is a starting is on the road to an organization be able to care for itself knowing that people transfer knowing that people's tenure in their teams and in their organizations is shorter than the applications let's just know that in advance right um and i think that's one of the benefits of the platform is a lot of that is now solved but as we've talked about the constantly updating security issues and the constant awareness that ci is as a first-class thing um we have to think about these things um hey it's good to have problems that aren't solved yet yeah well it keeps the the platform team busy also right someone's got to program the robots the robots are going to take over the world as long as they're thinking about their customers they're going to keep busy with stuff to do um and uh and then of course they're going to have those abandoned apps to to figure out but if they've done their job right from the get-go and they've worked with their app owners to build pipelines they should still be able to help make sure those abandoned we haven't even gotten to the other challenge yeah if you if you buy into the microservices um model of splitting out your application um i don't mean buy into it that it's wrong not to but you know the monolith versus microservice the moment you've got two apps that talk to each other how how what is what is that how do you make sure that your next version that's going down the pipeline could be deployed in production or is it dependent on something coming down in the other pipeline of the other app and that's just the two apps when you've got dozens of these things what is the the way in which you detach them all and feel confident that they can all be have do you need to how do i test my app against what's in production of all my dependencies and all the things that are dependent upon me um how do you bring that up i find these problems really interesting no i think we're going to have to save that problem for another day just looking at the time um we should definitely do a part two because i want that that's a good

one to explore because isn't the whole point of microservices is that you can develop your service independently of the others so you can you're not and it's yeah i think i think sometimes it's easy to do things um and only from that perspective do you realize that it's uh perhaps you should take five steps back off the cliff yeah it's just because you can stand on the cliff doesn't mean maybe you should um yeah so we're running close on time so i think we should uh give dr nick a couple minutes tell us a little bit more about stark and wayne for our listeners who may not be familiar with you guys so we especially very much around the pcf operational life cycle so we work with the product owner either to be part of their operational team to run pcf and pks and pcf um to uh or for those product owners we might you know own and operate everything so we'll be their sort of proxy team um and yeah i mean we love seeing this thing be a product we love apps coming on the platform and being successful it is just the nicest thing to see developers you know just be excited about the fact that something was fun and simple um it's fun finding out what they're trying to do and answer their questions um and uh and so then in that responsibility we liaise was the unspoken we i say unspoken we haven't spoken about them but the networking the infrastructure teams the the subsystems the email all those other things um it's super awkward and hard but it's the last time anyone has to do it because once it's integrated into the platform now everyone gets access to how do we send an email everyone has access to how do we access oracle no more jira tickets saying could i please have another oracle you know thing um so just getting an organization to that where people can self-serve is is the delight of our profession so yeah so we work mostly with uh a lot of pivotal you know uh directly with our customers helping you know customers have you know long success fantastic and um i think we we have

cf summit coming up we do philadelphia please come yeah uh there was gonna be when you're gonna be there we'll be there if you're a product owner you want to meet other product owners come along if you're an app developer and you want to learn more about deploying apps um you should come along you should also just go to youtube and watch all the talks that we've ever been given for the last five years yeah that'll only that'll keep you busy you should come to to the stocking room booth and say g'day i'll be floating around dr williams and the team and obviously there'll be a huge pivotal booth yeah i think so yeah we might be there it will be green i mean by we i mean jeff and i we'll be there but i don't know how much time we'll be at the booth right because you're going to perhaps do some podcasts we've got some other stuff yeah if you have interesting stories and you want to share on a podcast just like this you should uh ask yes just just show up doraine and i we should just set up in the middle of the show put your hand up yeah put your hand up and say i have an interesting story to tell right we could probably knock out i don't know a dozen podcasts over the the course of a couple days just just the people who listen to this one and then actually followed your advice unlike this podcast short and sharp is awesome yeah okay well cool so um folks will will come stalk you at sea of summit yep that sounds great um any other place where we're in SpringOne will be a SpringOne yes looking forward to october plan now uh buy your cowboy boots now by the way because it takes a little while to break them in and i just would hate to see people sliding around on brand new cowboy boots come october that's a good one piece of fashion advice for the for the episode i usually have something don't i that's very good just a little bit of context SpringOne is in austin this year which yes austin texas that's a cool little town have you ever seen all the bats that live underneath that bridge i mean not up close kind of famous you should go check it out at dusk it's really cool yeah i will i look forward to that you have no idea what i'm talking about google bats bridge austin and you'll see what i'm talking about yeah if you don't know what google is i'm

surprised that you've got to any of these podcasts yeah you've got bigger problems right you've clearly or you might have actually figured it all out and then just keep doing what you're doing there you go all right well thanks again and uh until next time thanks listeners

Tanzu Talk · 🍕 Get more pizza.