Building DIY platforms: now you’ve got two problems, with Matt Walburn

Episode 21 · 55:31 · SRT · archive.org · all Tanzu Talk

so you know we have some news items this uh this week it's the uh oracle open world slash java one and i forgot that oracle is one of those companies thankfully one of the few who like starts their conference on sunday usually on the weekend there's like and and not to uh not to put this down at all i mean it's obviously important but but in in my life as being an analyst and and uh interloping marketer or maybe it's the other way around like uh you know it's nice when events start on a monday even better when they start on tuesday but anyhow usually what they do on the weekend is maybe like a partner or alliances summit which which is fine but but oracle is just like pow let's let's put the head honcho out on sunday so everyone has to uh come in and cover the event it was uh it was a reminder that i haven't gone to a java one in a long time obviously yeah yeah that's you know sometimes that's nice to i guess get things started on weekends if you want to but yeah it's rough when you have to fly on the saturday and but yeah no they jumped right in on all the news yeah in fact in fact we were going to have a uh we were going to record two episodes this week but our mystery guest who will be revealed uh in a next episode this mystery guest number one couldn't make it because of because of java one but we have we have mystery guest number two let's unveil the mystery who are you uh yeah matt walden i'm a uh platform architect for our field organization here at pivotal and uh yeah thanks for having me sure you're you're uh you're one of the rare people also in the central time zone which is always very exciting we've got we've got to have a little herd mentality around that very much so uh speaking of of uh the the oracle oracle open world java one what uh what have you been seeing coming out of that richard yeah i mean as you mentioned it was uh definitely bam on the first day as tons of stuff was coming out with larry ellison out there doing his his ellison magic of you know hey now now cloud is getting started because oracle is here everybody relax now so you know put out some as usual big claims that you know now amazon has to

look on over their shoulder because there's a legit competitor out there they shipped some pretty good you know some good messaging around another generation cloud they've gone through a couple revs of this and uh lydia from gartner just a few minutes ago posted a good analysis of the new oracle cloud and again kind of called it a good mvp and doing a lot of the right things but they're doing a lot of things around infrastructure as a service they continue to invest in some of their software as a service sort of things whatever we're calling paths they're doing a lot of things with that so they're you know saying they're all cloud in thousands of engineers all working on cloud stuff re-platforming apps so a number of announcements we'll link to some of them in the show notes you know the biggest question is going to be is it too little too late is it a revenue protection sort of scenario that says if you're an oracle customer now this is the best place to stay with us are you actually going to get net new customers out of this offering i think that's that's what remains to be seen i'll have to go read the lydia thing is it is it behind the uh the the moat o'gartner or is it did she post it on her her blog oh very nice yeah she she blogged it so we'll uh we can add a link in the show notes as well to that as she's uh always has great insight yeah and and you know i think i think i think in past episodes we covered the uh the i as magic quadrant which which uh i i think i think you know lydia does good work uh hopefully that's a credible claim on my part since she doesn't really cover the stuff we're interested in so i don't it doesn't seem like i'm uh i'm plowing a field with a brown nose or anything but it it does see like her her coverage is generally good and and you like or i shouldn't say generally it is good and uh yeah i mean it's it's it's uh it's a tough thing to just suddenly sort of like sprightly and up and get into cloud stuff and and of course uh you know there's been a lot of a lot of entrances and re-entrances and into uh into cloud for oracle which is fine you know on on the other hand i often think that uh you know i have this this weird obsession with sun like sun had a tremendous amount of cloud stuff way back when and there's

actually a lot of uh cloud people who were uh uh i guess immigrants from sun stuff so so who knows all sorts of excitement there but but it does it does remind me uh back at one of my previous companies that i'm now back at again uh yeah i we i i worked in the part of the company that was that was trying to uh go go from a small public cloud presence to a a market leading cloud presence and it it one takes a long time and is extremely difficult so i don't know if you start the clock now or start the clock a few uh a few years ago or whatever but it takes some time to build that up but i think i think from from our perspective and you know an infrastructure layer right like i think one of the things that would be interesting would be to see uh you know if if we end up in uh if you can end up taking like something like cloud foundry or pivotal cloud foundry and running it on top of uh the oracle cloud which which yeah i think for they had a lot of interesting uh uh mud to sling at amazon about lock-in and things like that so that would be a nice practice and anti-lock-in i don't know what the opposite of lock-in is i guess uh libre free in something yeah i mean we uh you know i especially for our listeners who are pivotal customers i mean we'd love to hear if if this would be attractive to you that we often do do partnerships obviously based on where our customers want us to be and one value of running a pcf means if i love the oracle cloud and i want to be closer to my data services then you know pcf on oracle cloud isn't outlandish so we want to make sure we do provide some of that that uplift but we want to we want to hear from customers on that we're not just going to run out and do those things so no i mean look i live up here in seattle the word on the street up here is that they have pulled together a very high pedigree team of people to build this cloud this was not just hey let's grab some database people from internal and build a cloud they they recruited from the best shops up here so you know hey you're gonna give them a chance they've got some great talent they have clearly brand recognition and a a uh attentive customer base so that's uh let's keep

an eye on it now now matt to put you on the spot you you uh you you've been on the other side of the fence that is working at a at large organizations what uh sure did you guys have what did you what is a large enterprise you know to summarize yourself and talking with other people like what do you guys do with oracle and in large organization land um you know i mean i i have to i guess answer that from a historical perspective more than a current perspective but uh yeah yeah and you know i'm not i'm not looking for any gratuitous like hayton or whatever which which is popular with oracle but it's more of like i think i think a lot of people uh or oracle is good at like speaking about like the big grandiose stuff they do and i think it's always interesting to think about like in their huge portfolio all the not little but all the the components of the things that they do and how how how vital they are for a lot of people's it sure so i mean i think i mean specifically from the organization i was in um i think by and large it was largely their database right through all of the the versions and reversions and re-architectures of their core database product um would have been the lion's share of what we leveraged oracle for um as well as for some of their java runtime components right with with web logic as well as some of their identity management pieces but um i would say that it was it was kept pretty pretty in check and pretty much limited to those use cases yeah i mean i mean it seems like they're a uh they're one of the one of the few systems of record companies that exactly and and and they're few because all of the system of record companies have been gobbling up stuff for quite some time right like it's a uh that's a popular meal speaking of thanksgiving at least from years ago right i mean i'm sure Coté you remember from the years back gobbling up jd edwards and siebel yeah people soft and all these sort of just big things that all still i mean i don't remember which of those you can still buy right now i think a number of them but those are you know the hallmarks of most enterprise it shops and probably still sitting there i i remember running

into a uh a relative who i guess is what you would call a sales engineer uh i think i think he was kind of like an upper level sales engineer for jde as he would say and he had like two bookshelves this is a long time ago two bookshelves full of like three ring binders of like stuff was like that that was that was a fun time when you could sort of like quantify your knowledge based on three ring binders so there's an anecdote for you it's an important asset there you go so so also i don't know if it's related or if it's been going on but i didn't realize this that uh that that oracle is is moving net beans uh which which is the open source uh ide thingy uh over to the apache the asf the apache software foundation which which this reminded me of two things one eclipse and net beans like back when i was an analyst and a java developer that was that was they were head to head there i remember i remember when it was young Coté as a developer i used uh i think i used like borland first and then there was net beans and then eclipse came and that's all that was used way back then but then i also and and so it seems like i if i remember reading somewhere in in a right if there's like a million and a half users that they uh they they say they have which sounds great and uh but i remember one time i was in an eclipse con way back when when sun was independent and there was like a rival little net beans evening with a couple presentations like across the street you know the uh the uh the santa clara convention center and then behind it there's like some building that's named something really funny like behind the hyatt there so the person had rented out the lobby of that other building and there was a maryland monroe impersonator who was like getting people to come over and enjoy like little little weenies and spinach and and see presentations on net beans i i i should drag that picture up i think i even got a picture with her clearly your new uh your new profile picture on twitter that's right so if people wonder what analysts do i've just told you

peek behind the curtain cocktail weenies powerpoint and maryland monroe impersonators man that is your autobiography title anyhow i i think i think it it's uh yeah that that'll that'll be you know i i think it's it's in recent years like i used to think of like the asf as mostly infrastructure sort of software but they've they've been accepting a lot of like don't they have open office there too they've been accepting a lot of like front-end centric things yeah there's definitely a handful of things i you know for some reason dragging this back to pivotal there uh we we posted an interesting blog post this week on our approach to open source and what do you open source and what do you keep as ip and how do you manage all that so really interesting to see big companies as they continue to donate or give things to open source you wonder what goes on behind the scenes on those discussions i thought was a cool post by us it was our legal team talking about why open source is awesome but you know how did these big companies make these decisions now it's it's pretty fascinating yeah yeah i was just i was just noting that uh in twitter the other day and and because it's true that i i've encountered a lot of i guess you could call it open source objections i don't know if they're like speed humps or actual objections but uh i need to sit down and like sort through that i'm sure there's i'm sure there's some fun charts i can pull together that are basically like dude open source is okay don't freak out but it is it is uh something that i know is coming up a lot in the uh organizations that that we go talk with interesting so uh so also i think i think you read up more on this than i did it it looks like the uh java e8 is is is delayed a little while a little bit yeah i mean that's also part of java one as they're talking about that and last week it's kind of officially talked about that the goal is you know j8 by the end of 2017 and then the goal would be e9 about a year later which may be optimistic given that that eight slipped a bit but you know they're trying to do more cloud nativey stuff like they're going to talk more about it this week at at the java conference and you know it's just it becomes

tricky as you have these big stewards of things and oracle figuring out what to do with java you know the article we linked to on this of course i'm i'm happy to see someone comment saying that hey spring boot and cloud foundry is the best way of building cloud apps and java i think we agree that it's great to see you know jee eight nine or nine kind of tracking the spring framework five which we're already about to ship so you know we're trying to stay ahead of the game for people doing java microservices spring bring up being a premier choice you know on the jee side i think it's been frustrating for the community to see where is this going but i guess credit to oracle to try to to reinvest here and get a hard concrete plan forward but the community gets kind of stuck in this sit and wait period what am i supposed to do in the meantime we're talking you know 16 months out yeah till another version and a couple years until i get something that has modern microservices and this the basic things you already have in spring today no and and speaking of young me i mean this this made me think of two things one uh i think as far as i can remember not and and as far back i should say not as as comprehensively as far back as i as i can remember uh i think the java community has followed the uh the epigraph at the beginning of the mythical man month which is if good cooking takes time it's you know if you're made to wait it's it's to ensure that the food is good i messed that up but it's basically like you know hey if you want something that tastes good you need to wait until i cook it and stop complaining right like if you want if you want something quick i can make you a fried egg champ but it also like it was making me reflect and and this of course is like you know uh seems like really biased but man the spring framework has had like an amazing run of being like a huge source of innovation in the java world like i even back when uh when i was still a developer i remember a lot of the conversations we would have is like oh why don't they just do this and then you know annotations were added and all these i mean i don't pay attention to job anymore but it seems like there's a lot of uh interesting innovation that you can put into practice that

shows up in in spring and i'm sure many other places like i remember when java had this uh existential crisis about ruby for a long time and then it was you know that's an interesting historic example of like the java community not freaking out about i wouldn't call it a flash in a pan it's more like a long cooking thick chicken breast in a pan but eventually it gets removed and you're left with the pan so you got to live with yourself but they didn't get they didn't get too obsessed with uh with you know weighing around anyways yeah the long story is interesting yeah i mean if spring hadn't made a big change with spring boot and the like a couple years back i mean you wonder if there's a thesis that java would be seriously on the decline if boot hadn't boosted because all of a sudden now you are able to do things that don't feel like hey i'm being left behind by people doing things in dot net or node or go or whatever languages or people are kind of gravitating to is that all of a sudden java started to get some of these constructs whether it's coming up with reactive programming and all the things that boot gives you about look i can with eight lines of code turn up a web server because it's all embedded in the jar i mean i i'm a dot net node person and when i joined pivotal i committed to learning spring and boot and you know it's really pretty pretty great stuff and what we've been doing with spring cloud again we talked about a couple weeks ago this is really cutting edge stuff that's now baked into a java experience if that stuff hadn't come along you really wonder if we'd be talking about well i guess java's dead is it but instead you're seeing boot being downloaded you know millions and millions of times per month it just i i think we've done a good service hopefully then jee also kind of gives people some confidence that there's other things you can do with java besides spring but hopefully we can take some credit for keeping java in the spotlight so so then let let me let me ask you again matt when you go out including when you talk to yourself which you know there's not a sand of insanity it's fine you're just thinking sure to yourself like can you talk with like like enterprise architect types and other kind of people who are looking at the like long-term uh stability of their technology

choices are they do they have like a huge alignment towards like like stock java ee or are they interested in just like whatever works so how do how are people balancing that out sure so i mean i guess from what i've seen i mean there's a huge a huge embrace of of not just you know j2ee but but spring overall right i think for many of the reasons that i think pivotal you know talks about right just like removing a lot of that undifferentiated you know boilerplate code that that i think people you know don't have the time or the patience to write i mean i think spring is a great option there um and i would say that whether it's you know spring with java or dot net i i see a lot of enterprises still gravitating around those two those two languages and those two frameworks i mean just because of you know the longevity that they've had and the stability that they've had um and that doesn't mean they're not dabbling with some of the other things that that you know you and richard just mentioned but i still see a ton of momentum and a ton of excitement around dot net and spring well well speaking of excitement just just to close out before we get to this part so i think i think finally richard we both work in marketing our marketing job is basically going to be over over in the sense of like we will be able to accomplish all our hopes and dreams because there's no longer 140 character limit in twitter now this is slight slight technicality i think there's a a character limit but when you link to other things when you put urls in there including images that doesn't count for it so what this means now everyone knows there's two things that everyone on the internet knows one your video could never be longer than three seconds that's why there's only one listener to this podcast which is much longer than three seconds so you got to have short form uh multimedia as we used to call it two you got to add an image to a tweet or no one will ever look at it so now true now we can have maximum messaging and positioning in twitter right and we can add images no problem i think i think uh next next year it's gonna be it's gonna be done where we'll uh we'll all

be able to buy like english estates with all the excess money that we'll make because we'll be so successful with marketing oh so much time wasted now that i can do things like embedding gifts and uh videos and polls in 140 characters right all that's baked in so i mean look every tweet of mine will now have like a still photo of a bowl of fruit just because it can fit and it's fine it doesn't matter anymore so that's a but it's a good deal i mean again there's always these weird religious arguments about no it just has to hold this 140 characters and other people who want you to write you you know manifestos and no limits and you know at the end of the day i like the the forced brevity of tools like this twitter is a big part of my day-to-day experience in marketing and just learning things and meeting new people so i like that they evolve it i don't think they'll ever please remotely everybody but this feels like a good advance yeah i remember first reading about this in the summer and thinking if they if they did extend like if they really made it like you could do any length you want kind of like a facebook post then uh then then that would be kind of weird but like i like the idea of basically like you can have attachments essentially to your to your tweets so so that makes sense that'll be fun because it is nice to like attach images there's there's a weird corner case i always encounter which is let's see if you want to quote someone else's tweet you can't put an image in it which which is always annoying right i know especially when you want to include a funny yeah i want to throw a weird meme on it and i can't do it so exactly exactly there's a couple other things go ahead still coming that they are adding the uh you know Coté you probably know the quirk of doing the dot at symbol to try to actually start a tweet with a person's name but actually make sure everyone can see it they're doing away with that in the coming months and then some of these other things like hey any any other mentions don't count against your limit because you've probably been in these chats where 72 people get involved and you have six characters left to add any any text because there's no room left so yeah those things this is this is a good point a lot of people at the end of those those big twitter conversations which can i add those drive me crazy i hate those they they

they will they will find the exit door by saying i cannot have this conversation in 140 characters or or even you know the characters allowed by all you know i don't have enough characters now you can say like bro because it's always you know some person i guess you've got enough characters make the argument or you lose and i think i think that's that's how that's how stuff works in twitter if i remember it's very civil discourse all hot takes so speaking of hot takes i don't know what why i segwayed with that but hot takes is funny i you know i like to think we should do warm takes are there are there like ice cold takes i guess that's uh that's uh if you're doing quarterly publishing of a magazine someone must still do that like aws is super is really uh gonna take off that's a super cold take it's like good lord that's a six-year-old that's that's right open source is a trustworthy technology to use it's a super cold take super cold glacier anyhow welcome back matt you've been here a little while now now it's your segment why don't you why don't you uh give us a little bit of your background like beyond beyond being a platform architect and living in the central time zone uh as as we've we've established like what uh what would you say like is is one of your areas of interest at the moment like what what do you what do you what are you working on around here yeah sure so i guess i'd answer by saying that i i came to pivotal really just to explore the whole area of um you know structured cloud native slash platform as a service just in depth because you know i think pivotal's been doing that for a long time and has a lot of really compelling technology in that space um and that was interesting to me because i had been you know spending probably the better part of the last two and a half years at a large fortune 50 company uh building my own platform right so not leveraging cloud foundry or any of the great work that's out there from any number of other companies but really trying to engineer that whole that whole uh set of

capabilities um myself with the team that i was leading at my last company and and so let let's let's talk about the the idea of uh building your own platform a little bit i i was actually talking with one of my co-workers about about this yesterday how uh went on the team that i'm on where we go out to conferences and we traipse about and talk with people always there's a lot of traipsing like a lot of what you encounter are are exactly that people who build their own platforms and and first of all like like like what i mean people who've listened to this podcast kind of have a notion of what a platform is but like what is it what does it look like when when you're building your own platform like what are the components you're putting together and like what are you what are you doing like i i assume you're not writing a file system for example but right and yeah so i mean i guess you know my take on it i think is probably a little bit broader and um more comprehensive than i think maybe a lot of people go after but for me a platform is um you know something that i embarked on in order to provide you know a rich set of capabilities for more or less managing the full life cycle of not just applications that i was required to deploy but also all of the you know application dependencies infrastructure you know databases middleware runtime those sorts of things but be able to do that in a highly consistent highly automated highly predictable way um to manage all those things through their entire life cycle so to me that to me that's what what i would call a platform um yeah you you have a 30 words or less right you have a uh you there's like a little interview post that we have on the blog and if i remember there's you have a diagram in there of like all the different things that you need the uh the layers i need i keep forgetting to steal that for my stuff somewhere but but i i i think i think it's a good it's a good overview of like here's here's all the stuff that you need right right and you know i i try to break that down into you know various what i

call uh capability domains um and honestly i think if you look at it it represents you know all the different teams and silos of work that have kind of traditionally existed within you know infrastructure and operations and application teams but taking that to this idea of having an automated platform for delivering you know things like infrastructure things like all the you know operations and metrics and reporting tooling that you might need all of the tooling you need to have a very effective continuous integration and continuous delivery process in place as well as you know a lot of you know what i call the the application dependencies which are the things like application runtimes integration technologies database technologies etc and then kind of layered around all those things would be all of the different things that you need to keep all of those pieces um secure throughout throughout their life cycle so so when when you're like like in your experience and talking with people like who uh so i i mean the whole the whole thing of a platform initiative is is probably like at least a year long initiative from like you know uh let's say we come we come back from uh the the holiday vacation it's like what would it be like january 5th or 9th depending on the calendar and it's like hey we should have a platform well first i need to find time on everyone's calendar and i gotta schedule this out and then you know maybe a year later you're like boom we got a platform but like who like what's what's like the core team of people who are doing that like is it is it like a business side person or like is a bottoms up thing or architects or like what what drives this platform project sure i mean i think it can be driven in a couple of different ways um i think in my particular scenario it was more of a business um led initiative um i think the the main reason that we brought our platform into existence was to help migrate um a bunch of workloads for doing e-commerce and mobile and and like api type

services um from kind of traditional managed you know outsourced data center type hosting to running on a public cloud right and so in order to effectively you know take over all of those you know operations deployment support security all of those sorts of capabilities and to do that in a way that um where we would not get locked into a particular cloud provider it basically necessitated creating all of these different platform capabilities um in order to effectively move those workloads and be able to manage them in a meaningful way at scale you know throughout their life cycle so um so so do you think just to interject there do you think uh yeah do you think a common genesis if you will which makes me think about like spock and a bunch of white robes but whatever do you think i think a common genesis is like uh we're going we're cloudizing everything like is that a thing that kicks people off to to think about a platform uh you know i think i i don't know if it's like the thing you know that was definitely the scenario in my case right where we were trying to move these applications we were trying to move them into more of a you know loosely coupled microservices type architecture but i think honestly what i see most of the time is not that scenario um i think a lot of times what i'm seeing is almost like a grassroots movement within um you know operations or infrastructure and then occasionally sometimes within development right where they're trying to introduce you know honestly a lot of the the patterns and approaches and architectures that that i think pivotal embraces and um tries to enable within organizations they're very slowly like trying to enable those sorts of things as well and i think that's where they stumble across things like you know infrastructure is code and all the tools that are out there to do that um you know continuous integration and all the orchestration engines that are out there to do that and then i think now i think what what a lot of those same teams are seeing is oh well i think if i add containers and a container scheduler to that well that

that that that that gets to you know more of these outcomes and more of the kind of you know new school microservices automated ci approaches that i think a lot of companies are trying to embrace i mean when you're looking at some of the things you described there and some of it coming from kind of top down like business imperatives hey we want to go faster we want to have no lock in things like that but then the grassroots of gosh cd and ci are awesome how come everything's broken and you want to fix it both ways so you're seeing that kind of both pressure on both directions do you think if you were if a company's looking at cloud first and saying look let's just run this on azure aws google whatever yeah do they have the same feeling of trying to build everything from scratch versus using as a service components or does this come from wanting to have more control or keep it on premises sure where do you think you go into that extreme diy where you're actually setting up your own load balancers and and sands and and things like that do you like yeah what what takes you to that extreme um you know i think honestly a lot of that has to do with you know the culture of whatever organization you might be operating in um i think there's organizations that have you know very traditionally had a buy or kind of an embrace and extend sort of mentality um or like a let's buy but then customize um but i think there's a lot of organizations that either started from an opposite you know position where um you know call it an engineering culture call it a maker culture call it whatever you want but are maybe at the opposite extreme and then i think you've got a bunch of companies that maybe traditionally were you know more in the let's buy everything you know it isn't differentiating let's maybe you know use a lot of third-party solutions and and maybe even outsource a bunch of technology that have seen you know now that you know software is such an incredibly important component of of i guess what you would call like competitive advantage that they're trying to move back into maybe more of the middle where they're doing a lot of this engineering in order to you know create more of an engineering culture in order to maybe encourage um more engineering to

happen within their companies right so sure i guess i wouldn't say that there's any any one paradigm i think it just all depends on you know the culture of the company and where they're starting from yeah that makes sense i mean do you see there was a an interesting tweet last night uh i i saw from actually lydia and funny enough from uh gardner it was uh recounting a bank that had gone through kind of this is the the new modern tech stack they built to run their the bank technology and interesting post really cool stuff it was exactly the things you've described here in terms of the platform components and someone i respected this morning had commented on it saying this feels like kind of like a premature optimization so how do you when you were even doing this yourself and i mean did you have a pretty schematic all those things you were going to have to do or does it grow organically as you come up with more needs i mean can you accidentally plan way too much up front without knowing what you're doing yeah i think i think you absolutely can um and i would say that i was more of the latter right where i actually did spend probably the better part of i mean this is going to sound obnoxious and it probably is but the better part of a year you know strategizing architecting come up you know coming up with what i thought was you know probably like the perfect you know reference architecture or capability architecture for all the things that i felt that we were going to need to effectively manage you know a pretty significant e-commerce site running on public cloud right so i did have the schematic but i think i think that probably most people don't start from there honestly they probably don't have the luxury of that kind of time on their hands in order to invest in in designing something like that up front and so i do think they probably get more into the smaller iterations where you know maybe you're embracing a you know container technology such as docker right and it's you know giving you a lot of those quick wins where you know you're able to get your code into into production maybe a little bit more quickly with a little bit more consistency but then you know a lot of the a lot of the things that you know i talk about are okay well what happens after that as part of you know running in production on day two right which is

i think where a lot of the other components start to come into play yeah that's something um from the uh the interview you had that uh i don't know i i don't think i thought about it in that much detail when it comes to building your own platform like usually the quip is like you know the quip that i and other people use is like you know now you have two problems you have you have like the uh the platform you have to maintain and also the application that you want to put on top of it so instead of one you have two which you know sounds funny um but the the detail on that is that um to the to the quippiness of that like you do have a product that requires all of its own process and product management and ongoing work right like you don't you can't really like just uh throw up a wiki page and hope that it gets maintained like it's it's right it's as real a product if not realer as the application you're going to build on top of it because it'll be supporting multiple applications so it's almost a uh a complete parallel process that you have to do now i i think i think the pushback that a lot of people would give is no it's not that like yeah that that like the the uh the application team or whatever can can like take care of it but i mean i mean and and to that end can can you kind of like quantify i mean is it like is it like a doubling of resources that you need or like what there there must be some part some point where like if if you have one application and one team or maybe and they have two applications like it seems like there there are cases because we see this demonstrated where you can build your own platform but there seems to be a certain like scale that you achieve or where you don't have the staff for it where it's not possible so you know we've been having this ongoing discussion about roi stuff which is as as people who read all the stuff i do knows is like this albatross on my back but like it seems like it seems like that's where a lot of the cases is like how much do we need to spend resource wise on building our own platform

versus not doing it and so how do you like quantify what that effort is it's like longest question ever there yeah i mean no no no and i i i struggle with that same albatross honestly because i do think you know when you're getting started i think that that the the the challenge or the problem of like doing this whole platform thing i think on the surface looks really small right and i think that's why it's really attractive you know individual development teams individual developers to take on some of the burden of managing that right because um there's just a a level of scale that they have haven't reached yet but i think what i would say is that even even if you're a single team or a small development team and you're currently handling all these platform tasks like at some point you know all of the same life cycle and management and product management and all these things um i think eventually come come home to roost right because if you're going to continue to justify having a platform at all like in some sense it has to be competitive with what other platforms are doing right because otherwise why would you have your own platform right so it's one one thing to get a form built once i think it's another thing to keep it going because i think by its very nature if if you're going to continue to justify having a platform like it has to be you know it has to achieve some level of feature parity um and stability and and serviceability as you might find with some of the more commercial offerings and so i think that's where you get into um not just deploying the platform the first time right like not just standing up kubernetes or whatever you're using but actually okay so if i'm going to continue to be competitive if i'm going to continue to have this platform it's got to do more over time right so then you're into you know planning for you know what are the new features i need what what what new releases do i need to upgrade to what are all the patches i need to do um how do i introduce you know changes to my various platform components without causing a lot of disruption right and so that starts to look like well now i'm doing things like uh it's not a great analogy but

test-driven development for my platform components and doing things like continuous integration for my platform components and so i think over time especially as you start to have more and more consumers the the the expectation and the and the demand on that platform team goes up um in a lot of the ways that i just that i just mentioned right and as you continue to grow the platform and it continues to have more capabilities and more moving pieces um that starts to look like a big snowball running down a pretty big hill right so you've kind of become our uh i think our accidental closer when it comes into deals deals it's like yeah oh god this is a big ball of mud customer who just has this platform that they know and love or they're clinging to and we really want to talk to them about pivotal cloud foundry and platform so if we're talking details when you go talk to a customer sure how do you help them decide whether to either keep going or start with a role their own or the target let's say pivotal cloud foundry and kind of second part of that what are those things in the platform that you talk about that say these are things you don't want to worry about sure so that's yeah let me uh there's a few questions in there i think so maybe i'll i'll start at the beginning and then maybe just help guide me through a little bit um i think i mean by and large what i try to talk to people about especially you know large companies because honestly some of these problems are kind of like big company problems more so than small company problems um but that if you're really going to grow a platform and have it be something that's not just providing service for you know one team or two teams but for your entire enterprise um then there's a certain you know scale and size and maturity that that platform has to have and so what i end up doing is walking them through um basically the journey that i went on over that two and a half year time period from not having a platform to having a fairly feature complete comprehensive platform that that in a lot of ways um had a lot of the same features and functionality that that

cloud foundry has um and so i walked them through like all of the resource but actually even before the resources just walking them through things like the time frame and how it actually took you know close to two years to get to a feature complete platform um i also walked them through all of the things that they might not be thinking about from an overall capability perspective right because well you know i think a lot of people like very quickly gravitate towards hey i'm going to build a docker container on my laptop and then i'm just going to deploy that to a kubernetes or other scheduler in production right and they don't necessarily think about you know how i continue to manage the the quality of those artifacts that i'm pushing how do i continue to you know upgrade them patch them maintain them how do i provide for you know seamless deployments of those artifacts how do i do monitoring how do i do reporting how do i manage um particularly in the case of like public cloud deployments how do i manage my cost on an ongoing basis so that i don't have just you know over deployment of of these these these containerized or virtualized resources um walking through all the security things and just walk them through all these components and then talk about you know each one of these components has a life cycle right like if you're really going to have this thing survive and scale and be sustainable over time like you can't just be thinking about how do i deploy it it's how do i actually manage it from from cradle to grave right and i think i think that starts to open people's eyes quite a bit um and so that's just kind of from a capability perspective i also walked them through you know just basically what what i did from a resourcing perspective right so if you take all of these different capabilities um well ultimately they need to map back to a set of engineers right and so at least in the the way that i tackled the problem i actually stood up uh product teams around each one of the capability domains that that that factored into that overall platform so i had a product team for you know the

automated delivery of infrastructure i had a product team for all of the automated codified ways that we would do things like monitoring and instrumentation um a team around continuous delivery a team around middleware a team around data and a team around security right and so it's like if you're going to manage things for an enterprise at scale throughout their life cycle it just it it takes a lot of people right and so i end up showing them that hey after i was all said and done like i had this amazing platform and it was you know absolutely the most the most rewarding most um fun most intense period of my career but at the end of it i had this platform um that was kind of going to be an ongoing concern for the company i had a team of you know about 55 engineers that was having to maintain it um and if you put those two things together there's there's a pretty hefty price tag that you end up associating with that effort and so the question then then ends up being like hey if you've got you know in my case it was you know around between seven and eight million dollars of payroll that i was spending to engineer and keep the lights on for this platform right and if you've got that kind of money to invest in talent and engineering do you want to be doing that for a platform or do you want to be doing that for your core business right and so now you have i think those are right so those right so those are the two things right yeah so well i think you've got three problems right you've got these two ongoing concerns right so you've got the platform problem that you're trying to solve and the and the overall business problems you're trying to solve and the third problem is that you know you're robbing you're robbing the business to run this platform in a sense right and so it's it's illuminating all of those different things and then i think also illuminating that um you know when i started there there weren't comprehensive solutions out there you know like cloud foundry or or a number of our competitors so like building a platform was something we almost had to do at that point in time but then there's much more there's much more viable options out there where you can remove a lot of that you know undifferentiated engineering that my team had to do so that you don't have to be incurring

that that cost or those those delays um yeah i mean that's that's that's that's one of the the sort of like uh theoretic footnotes or or whatever that i have in this discussion always is that uh you one reads a lot about people who have built their own platforms because you know if you if you reverse out the math i don't know what i said there but but if you sort of walk back from presenting now it probably means like two to three years ago they started out doing this and and at the time the the uh i don't know what the new phrase for off the shelf is but the off the internet like options were not as viable as they are nowadays so like it would make sense that people had to build things just as you were explaining but nowadays uh if you weren't in that little that little vanguard there's a lot of other options that you have so i i had i don't know i don't know if richard has more but i had just one more one more question which is um you know you know in in in the breadth of of all the things you need to support one thing that i think is interestingly different about uh about the the platform as a service or the platform area nowadays is a much tighter integration with with continuous integration and continuous delivery like all of those pipelines which you mentioned many times right like i think right i think i think aside from the buy everything from one vendor pitches that you would see in the 2000s like people tended to keep those things separate and so so i wonder like how how integrated do you think your your cicd stuff is into your platform consideration like how how much are those things tied together or should um together well i think i think you absolutely need to have a platform that supports whatever you're trying to do from a cicd perspective right so i i think there has to be um you know a really great integration between you know the practices and the approaches that you're trying to use to enable your software development organization um those have to be enabled and present

within your platform as well right and and the way and honestly that was one of the big one of the bigger of the many kind of guiding principles that that we took when building that platform was that we wanted to be able to continuously integrate and deliver all of our platform capabilities so that we could really not so that not only could we you reap all the benefits of you know better quality and faster time to market and all those good things that we always talk about with ci but so that we could also really intimately um understand and identify and empathize with the developers who ultimately we're going to be consuming right platform right right so that so that we could provide a very clear link between um the practices that that we're trying to enable because we're we're actually using those practices too yeah no that's interesting i mean i i i i picked up on that one little part and wanted to like pull on that string because with uh there's been several several organizations i've spoken with recently who spend a lot of time perfecting their pipeline which which makes sense right like like there's a lot of of uh i mean that's the software you're building it's it's it's your workbench in in some way and and i i've been curious to like see uh you know sometimes i encounter people who like they give you an overview of what they're doing and it's sort of like gold plating to some extent or maybe silver plating but but i think i think over the next few years that'll be you know the cicd spot will be an interesting area to look at in the same way that as we've been talking about it's generally a good idea to get a platform instead of build one i wonder how much of like the pipeline process if you will not the actual tools itself but the actual kind of process that you do will be sort of people adopting conventions that are out there rather than coming up with their own things yeah i mean i think that gets gets back to like where is value actually added right and i don't think probably the value is added in building you know a very super custom unique pipeline like the like the value is probably added in how how good are your tests right so it's it's i think it's

kind of like you know we talk about these value line conversations at pivotal a lot and i think you can have value line conversations you know within platform within development within within ci and cd um yeah i think i think you're right on with the silver plating comments yeah so yeah a couple last things for you so sure when you do talk about this i mean whether it's the pipeline or others when you talk about to platform builders what are the pcf features that click the most like where do you see light bulb moments when you're talking to people who have been building platforms or are thinking of going down that journey are there a couple things that when you show x or talk about y that they go sheesh i don't want to have to do that myself like are there some of those uh yeah i think i think definitely right and i think a lot of them are uh i think what i would describe as all the infrastructure is code that you no longer have to write right which which kind of cuts both ways right because it is a lot of work you know whether you're using shaft or puppet or any of the other tools that are out there it's a lot of work to write and maintain and continue to grow like your base of infrastructure code um and it's also really hard to coerce the organization into using you know a set of canonical artifacts around infrastructure code and i think that that's one of the things that i think pivot that the cloud foundry product makes really easy in both just an overall execution perspective like in like the actual getting the the automated config managed you know um infrastructure built out but then also from a governance perspective where it's it's much easier to to manage the the automation and and the structure within the platform than it is through you know github repositories that might be getting cloned and forked all over the place so i think that's that's a big piece you know whether you're talking about the virtualized infrastructure management we do with bosch or with the actual um container build and deploy process that we do with our build packs and elastic runtime i think those are two things where like like all that effort

that that you may be spending right now is just gone right so that you can focus on some of the more higher value things like you know helping helping customers actually you know re-architect their applications effectively use the platform um getting into things like performance and availability some of those things instead of you know mucking around trying to perfect the perfect chef cookbook yeah awesome yeah so i want to close with uh one of my favorite hot takes from the last few months is something you wrote in an internal chat and i saw you write it into another paper so clearly you like the line but okay so what you said was i'll read verbatim the uh the over empowered full stack developer is an over correction created as a necessary response to the inefficient processes and poor outcomes created by yesterday's suboptimal ops teams so i think that is a bold and interesting thesis that sure to some extent this idea of a full stack engineer is not a desired end state instead it's the kind of messy middle we're in as as a result as you say because ops just there wasn't the right relationship between dev and ops so you know i want to give you a minute to maybe explain that but then also what do you think is the desired end state a full stack engineer is an over correction what what is the right place to end up yeah so and i know that's kind of a product provocative bold statement um but i think just coming coming from my background i there that for quite a while we we really did embrace and try to propagate this whole idea of the full stack engineer right like whether that's in um and or operations and infrastructure or in development um and i think what we saw was and i'll just focus on development for for a second is that during the era in which we didn't have api driven infrastructure right like we didn't have true infrastructure as a service we didn't have um platforms that provided the services that we really needed um and expose them in a way that was easily

consumable in a self-service way to developers i think that the over correction to that was this idea that development teams along with actually writing kind of like the the the the added the value added business functionality that they have within their applications were also spending a ton of time doing things like writing chef cookbooks and using you know uh puppet or chef or whatever to automate the deployment of not only their application but also all of its dependencies and infrastructure and i just saw um that was a very very very exciting i think i think time that we were in and i absolutely helped to support that and drive that forward um and the whole idea of really empowering um developers and engineers to to own their whole stack but i think in hindsight you know having come through this platform experience and and now with a lot of what i'm seeing at pivotal customers i think i would say it's an overcorrection because ideally you wouldn't have developers focusing on things that aren't applications right you would have them focusing their time there and instead having a very rich um feature complete set of capabilities being provided to them in a way that's super easy to consume right so that they don't even have to worry about those things those things are present um and available as as much as like turning on your water faucet or picking up the phone and hearing a dial tone right so i think that's that's really what i'm talking about is that there was a gap between let's just say like enterprise um infrastructure 1.0 and the developers that forced them into this model of having to do everything themselves because they weren't getting what they needed i think with you know everything that we've seen with cloud and with cloud native platforms and those sorts of those those those approaches that have developed over the last couple of years i don't think for most people it's probably necessary to engineer the full stack anymore hmm yeah i i mean this is this is the benefit and uh what's the opposite of a benefit problem with

developers is is uh they code their way out of any problem so right and and you know i i remember uh to speak fondly of myself like quipping that i think i think many advances in it are based on the fact that someone didn't give the developers the credentials right like so the the server admins or the security people or whoever like nope you you can't like deploy vms or whatever and so boom you write your own platform which is right it just it's just an example of people coding their way out of a problem and and you know also uh before we wrap up i should add if you really want to be a platform developer we're hiring you should come work for us and you can that's it should be building platforms right people who actually need to build platforms exactly there's lots of people who are hiring if if what you want to do is be more of a uh i wouldn't say an uppercase systems programmer like work on uh like hardware and stuff but if you want to work on platforms there's many options out there so uh so hey thanks for being a guest it was it was good thanks for having me so if people want to uh follow on with you and see if you have any uh whatever the temperature of the take may be see see what else is going on what what should they what should they do you got twitter or a blog or what what's what's what's your what's your thing yeah you know i'm i'm out there on twitter at matt walburn so pretty simple first name last name uh i'm out there well there you go so you can you can look that up well as always this has been Pivotal Conversations if you want to look up the show notes you can go to pivotal.io podcast and we're in soundcloud as well that's our our not so secret back end you can go to soundcloud.com Pivotal Conversations and uh if if if you like this podcast and and you want to help out in a way that costs nothing it's good to subscribe to the feed just put that in your old uh your old podcast listener i use overcast myself but i understand there all are alternatives out there also it'd be great if you go into the itunes store and leave a comment and it's even better if you tell your friends that they should listen to this so so we can get a broader reach for all of our uh

all of our delightful nonsensicalness as as i like to think of it and with that uh we'll see everyone next time bye bye

Tanzu Talk · 🍕 Get more pizza.