December 21, 2022
Season 1, Episode 5
You wouldn’t build your own car, so why would you build your own platform?
Jon and Jay consider the pros and cons of building your own. Spoiler alert: they advise against it. Cloud provides innovation behind the scenes, whereas building your own means you have to hire a team and sink enormous costs before providing any business value.
In This Episode, You Will Learn:
Themes Covered in the Podcast:
Quick Takeaways:
Follow for more:
Jon Shanks: LinkedIn
Jay Keshur: LinkedIn
Jon & Jay’s startup: Appvia
0:00now on necessarily the same problems that we solved back then
0:06um but in terms of like outcomes that you’re trying to drive for development teams
0:12it’s still quick time to onboard in a standard way I think it’s still being
0:19able to iterate
0:24hello and welcome to another episode of cloud unplugged um it’s season two episode seven
0:32um and I’m Jacob Shaw I am John Shanks and today we’re going to talk about uh
0:37building Cloud platforms versus buying them and all the different um kind of pros and cons to doing that
0:44so in the last episode we talked about um you know building internal teams that
0:50focus on um providing a function if you will um to to the teams that are actually
0:57building on the the business problem that they’re trying to solve of external
1:02organizations um now we’ve obviously we’re in a in the software delivery World we’re in this
1:08situation where um internal functions are kind of created and stood up to help people that
1:16are delivering actual business value I mean that’s kind of the definition of devops to a certain degree and how that
1:22discipline came about um they’re there to surface or to help the developers
1:28um and how that sort of evolved over time means that there’s this new sort of
1:34concept of platform engineering that has come about right yeah
1:39um and well Cloud platforms and then platform engineering and how that’s sort of evolved over time as well
1:45um so I thought we’d uh we’d sit down and talk about um kind of the pros and cons of of
1:51building a platform the types of things that you would think about and then um you know touching on some of the
1:57experience that you have of leading teams that do exactly this so
2:02um I guess where should we start um maybe in terms of the time
2:08um taken to sort of build a platform or to even know the problem that you’re
2:13going to solve as a platform team um like when when you’ve started in in the places that you’ve worked how did
2:21you go about even aligning on a direction or a vision of the problems that you’re going to solve
2:27[Music] well [Laughter]
2:33I guess a lot of the platform stuff historically was from experience of I
2:40guess historic models where I come from companies where we’ve done embedded devops and so I kind of seen that
2:47through in other organizations or previous organizations working for and
2:52then could see that it didn’t wasn’t very optimal and wasn’t very efficient also some of the technologies that
2:59existed back then were quite um I guess arduous and not as
3:05commoditized as they should have been for software delivery yeah things like config management and stuff like that
3:11it’s like not like whole configurations of machines VMS or physical machines
3:17from you know bootstrapping and then configuration and then the app yeah exactly right and all the other stuff so
3:23it’s very long-winded before you even got to the app um so I guess from a vision perspective if
3:31you look at those then there were already existing issues that you’re looking at and being like well this could be better
3:37um so if I was to do it again maybe embedding isn’t the right answer because people aren’t consistent
3:43um you don’t get consistency through people we can kind of see see that it’s not really a community that you’re driving either right so you’ve got like
3:49lots of kind of um disaggregated ways of building something but you you’d think
3:56that people would maybe want to talk to each other if they’re building something so that and you you almost expect it to
4:03happen because they work in the same organization but it doesn’t not as much people do share yeah I guess it’s
4:09probably too too again we’re always into a bit philosophical with this but um
4:15people tend to only share the successes the good things yeah and when they share things so it’s usually like they’ll tell
4:21you about now why it’s really good but just done but they won’t tell you about necessarily why like while they’re
4:27failing all the failings going on before they got to the good thing so there’s obviously that bit humanistically obviously people don’t do it but second
4:33thing I think is that um your time pressured and you’re in a team even when you choose tools there’s a
4:40thing I found really interesting was even with tools that you choose again tools not products which I kind of have
4:46a bit of an issue with sometimes but what’s the difference between tools not products I think tools lend themselves
4:52to being exceptionally flexible okay um and products tend to be a bit more
4:58prescriptive on the outcome they’re striving for and and more I’d say a product’s more focused on experience
5:04um and efficiency and a tool is more there to do a very Point thing in a very generic way I feel like a tool is just
5:12like a single point solution for a thing and then a product generally has more uh
5:17things that it’s trying to solve as a as a like a definition um of it really yeah they do cut very
5:23Niche products that only do do a very specific thing um but I think it’s just more a tool is
5:30something that is more flexible I would say tools more configurable as in like to do lots of things like is in would
5:37you say like terraform as an example is it a tool or a product definitely a tool really not a product interesting the SAS
5:42bit of it will be a product okay but the tool itself is a tool it’s not a product fine
5:48um a product for me is like something that brings an outcome Jenkins taller products
5:54that’s a tricky one actually it’s like at all guys as a product
6:00oh that tricky Butler tricky Jenkins
6:05um yeah I’d say that was more of a tall tall experience because it’s pretty much it’s so flexible I think that’s the
6:11thing I think when there isn’t a defined outcome it’s striving for you can basically do loads and loads of different things for me that’s very
6:17Tully um basically what we’ve um uh what we’ve aligned on is that no
6:24one knows it’s just a taller and product who knows no joke we don’t know yeah no
6:30one knows yeah basically we just don’t know a big difference there but yeah maybe
6:36next time you’ll you’ll you know have an opinion that is based on you well too I mean I don’t know I just think in terms
6:42of like um any tool in the industry in any
6:47different industry is I guess maybe synonymous but I think it’s more you’d
6:53need lots of different tools to then create the outcome if you see what I mean so then I think about a job it’s
6:58like you need a variety of different tools like if you’re any type of Industry to get the outcome
7:04honestly I think they’re like they’re semantics again because if you think about it
7:09um let’s say a hammer right um the person that’s using that hammer
7:15is like oh yeah give me that tool give me that hammer but the company that’s building it are like oh here are all my
7:20products that I’ve built so it honestly I think it’s which side of the fence you’re sitting on and how you think
7:26about the thing that you’ve delivered um there might be even like product management that goes into building that
7:33thing um does that make it a product like is it more about how it’s built than the thing that’s solving who I don’t really
7:40know I think it’s just again just semantics so do you think though a tool
7:45is more for a person and a product more for a business probably imagine you do get b2c products so there
7:52you go um anyway yeah um
7:57you know it’s just such a weird what what were you asking um what was the question what was the question so
8:03outside of tools outside of tools and products um what was the problem that
8:10um you know when you’ve built a cloud platform um how did you know which are the problems to solve and then how did you
8:17go about doing it um well I guess I didn’t necessarily know all problems to solve obviously at
8:25the beginning but I definitely knew what problems you would would be introduced again that’s what I was saying about I
8:30mean in the previous episode like history tells you a lot a lot of stuff right when you’ve been through the motions many times over in lots of
8:36different companies you see things repeat and so you then like well there’s no poke making things that you know
8:42there weren’t necessarily mistakes but on reflection hindsight’s 2020 you look back and you think well actually let’s
8:47do this again I wouldn’t do it this way I think that’s literally exactly what munzo turned around and said we said
8:53yeah yeah but I guess that’s the point isn’t it is that if you do things again yeah would you do it the same way no and
8:59that’s experience that tells you that um and I guess you’ve watched things go about in certain ways but
9:08that you know like normally you wouldn’t necessarily do things a certain way
9:13um and there’s always something that you would change because that’s the whole point right you’ve learned that
9:19something doesn’t work and now no it doesn’t work it’s just that it was less
9:24efficient and it could yeah because you’ve gone on the journey already exactly yeah so it’s like that it’s like anything’s like well that was a really
9:30long route to get here you know I mean like actually now on hindsight I could have probably got here in 10 minutes much cheaper yeah exactly
9:37um but yeah I guess that’s the point so you kind of look at the limitations of that was it tools I was saying I’m like
9:43when you’ve introduced things people standardize when people used to standardize it standardize under Tech you end up with like these Tech kind of
9:50Radars or Tech maps of the like hey this is all the tech tools that we’re going to use and you’re like yeah but you can
9:55use them in a million different ways exactly you can use them in so many different ways and that’s exactly what
10:01would happen yeah and so you’d end up with like variations of pretty much the same thing
10:07all slightly different and when you look at that you’re like well this on
10:12aggregate over time is going to be expensive because there’s so much variability going on now no
10:18standardization other than one tool yeah and insecure like people exactly that’s a good point yeah yeah people don’t talk
10:25necessarily about you know how insecure having all of this
10:31um you know even if you want to talk about autonomy of choice or whatever but when you’re introducing so many
10:36different moving parts and you’ve then got to configure those parts and those those configurations aren’t in any way
10:44um defined as sort of a practice that is then you know repeated and reused that
10:51is that’s a lot of risk it’s a huge amount of risk I think that’s where the biggest the biggest constraint I mean
10:56other than inefficiency obviously it’s not very optimal to be kind of solving the same problem in
11:03different ways multiple times over across different projects but then also because of that then yeah that’s it now
11:08you’re like well I can’t actually guarantee that we’re in a secure place now because there’s so much variability so I kind of have to know but they’re
11:14the limitations so you kind of looking like well actually let’s not do that again yeah right because that’s like it
11:19won’t scale nicely for the business because you’re going to introduce risk um and also it’s not efficient and
11:26people become very project focused and that’s not necessarily a bad thing but if you’re trying to be more business
11:35focused um overall more holistic than you want teams to be sharing all the
11:42time about everything going on across projects because it’s then you do get to see the patterns much quicker yeah than
11:48like being disaggregated and maybe they’re going to be more projects focused which means they become more
11:53part of the project team and less part of maybe the function of like how do we solve these challenges across all these
11:59projects in a consistent way um so it kind of switches you know
12:05you’re more like you don’t see all those problems because you’re not part of it you’re just focused on getting the project over the line it’s funny right
12:10because then you also have you know the concept of like Enterprise architecture and if that’s kind of the problem that
12:18they’re supposed to solve really is um seeing all of these patterns that um
12:23are being developed on into different projects and because they are almost a
12:29central function being able to kind of connect the dots but that’s not normally how like anywhere that I’ve been that
12:36has that government yeah exactly is that when you centralize just doesn’t really work does it they’re too disconnected to
12:42pretend they know what people want but they really don’t know what people want in the end yeah a centralization of power without necessarily yeah you’ve
12:48got to be living it day to day I think and when you’ve got people that are trying to deliver the projects but from
12:53a central perspective then you’re kind of sharing because you’re all in it together on making the project
12:59sufficient um all projects not just the one you’re working on yeah and I think that’s a
13:05nicer thing because you get to see the patterns and then you can share problems and find solutions to those problems
13:10faster that’s the benefit of having platform engineering the difference I think now
13:16is when I think maybe we’re in another kind of period but back then containers were
13:23just new so obviously lxc and you know um the changes in the kernel made by
13:29Google it’s Colonel but for c groups then Docker came out but it was in its infancy
13:35um so that felt better as a package management system for applications than say using config management to then
13:42deploy an app onto a VM yeah much more repeatable and easier for developers and
13:49can obviously test it locally um so there’s like added value there on delivery and then kubernetes came out it
13:57wasn’t out when we started then it kind of came out yeah 0.2 is miso Samaritan
14:03and
14:12um but yeah so there was like then like how do you schedule the running the containers and then obviously then swarm
14:19kind of came out and then kubernetes had already come out and we looked at that when it was 0.2 so it’d just come out and there’s loads of things missing
14:26um but then you could look at their road map and you’d look at some of their proposals and you’d read through and how they’d conceptually modeled out
14:34um everything made more sense right so you’re like oh yeah actually a replica of something and then it became a
14:39deployment later and um storage and you know State um and so they thought through there’s a
14:47really good like way of flux yeah that thought conceptually well about the problem before that even engineered yeah
14:53anything for it and that that’s what I got bought into but that felt much more optimal for software
15:01delivery for because of those paradigms so then you’re like okay well we can obviously build on this to form a
15:07platform team you know which will then give a consistent way of all the T all
15:13the projects delivering but obviously that was hard back then but now you’re like actually there’s
15:19loads of things out there that kind of do that and you also don’t you’re not building well unless you’re kind of
15:24on-prem or whatever you’re probably not building kubernetes from scratch like the cloud providers exist right well now
15:30yeah exactly so kubernetes you know is almost commonly seen as a platform for
15:35building platforms but when you’re getting it commodity and what’s the what is the next problem to solve um
15:42what are you getting commodity and how what’s that sort of developer experience look like
15:48what do you mean so um so you know we we based a
15:53um a platform we built a platform on using kubernetes as a main kind of
16:00component of how teams were delivering software um but there are a couple of things in
16:07that and that we could probably talk about but I’ll just touch on so it was like a multi-tenanted platform which
16:13meant that uh the time for a team to get started because you weren’t reinvented
16:19you know because you didn’t have to there was already something that exists and that was just scaling and that was
16:25kind of been built in the way that the the the the um clusters were built and stuff rather
16:30than creating new and creating isolation which is kind of commonly seen as best practice now so there’s different it’s
16:39almost like like you were saying the problem has moved to a different place um because there’s so much
16:46um uh Innovation that’s gone on with the cloud providers that they’re offering up
16:52a different way of consuming a base level platform yeah so the problems to solve now aren’t
17:00necessarily of the same problems that we solved back then um
17:05but in terms of like outcomes that you’re trying to drive for development
17:11teams it’s still quick time to onboard in a standard way
17:16it’s still being able to iterate on your development on your on your products in
17:23a way that simplifies and abstracts a lot of the details so that you don’t
17:28have to get into the ins and outs of how you’re managing the infrastructure or the networking and all of that stuff um
17:35but it’s but people still kind of think that they
17:40you know there’s it’s Unique that they have to you know they they’re like okay uh instead of
17:47focusing on a product that solves those problems I’m gonna solve it internally
17:52for my business because my business is unique um what why yeah I wouldn’t say that’s
17:58true I don’t think people I don’t think people set out to try and
18:04take on that big problem I think people start like the bit the bit that kind of
18:10goes on which is the same as Cloud it’s until you’ve it’s very easy when you’ve gone on the journey as many times over
18:16to kind of understand the problems better yeah um you know and where you end up when you make decisions because you make
18:23choices and choices of ramifications and you’re not quite sure what those choices mean in the end until later and
18:30obviously that’s hindsight right because you’re like oh actually maybe that wasn’t the best decision though yeah but
18:35you only know that through basically rehearsals right you’ve gone through it a few times before or that you kind of then learn from those things I’m saying
18:41experience counts for most stuff it’s hard to replace that in truth but I
18:46think the other thing just to pick up what you said is I think there’s a there’s a few opinions I have on this
18:52because the cloud to me is the platform that’s the first thing I would probably point out a platform
18:59if you’re all in on cloud yeah that’s that’s it that’s the platform right if you don’t have on-prem and other stuff
19:05or that could be now you’re kind of cloud on-prem but yeah well if you’ve got multiple clouds again no it’s the cloud the cloud but
19:12the there’s no I mean there are obviously um Solutions out there that kind of standardize across clouds
19:20um but let’s say you’re on Google and AWS yeah right um do you have two platforms do you have
19:27one Cloud platform like well I think that’s a different but if you I guess
19:33the first thing is if you’re winding packages from just just making if you accept the fact that the cloud is the
19:38platform sure right it’s giving you the components it doesn’t mean it’s the entirety of everything okay because
19:44obviously you might want to use something from some of the cloud types I’m not saying that it’s just a cloud
19:50vendor or that it’s just the major Cloud vendors because you might use something else like GitHub wasn’t owned by
19:56Microsoft originally right so you might have had GitHub that again is part of the cloud I would say that’s what I’m
20:02saying is okay it’s a bit more generic from just maybe the car but there’s loads from a cloud vendor one of them
20:08you get loads yeah and then you get more even from another Cloud vendor and you should be able to choose maybe because there’s cost reasons for one or the
20:15other but then when that decision has been made on like because that’s a decision you have to make do I do I have
20:22a mandate because I have to do multi-cloud or is there value from a commercial perspective on being able to
20:27pick the right tool for the right job um in the business and actually that’s more economical
20:34commercially and it creates good commercial competition internally to the business having the cloud vendors maybe I’ll get a better deal yeah right
20:41that’s a decision though then you then drive strategy from that because if those things are important then you’re
20:46like okay well now how do we make this work um and I think you can get into platform
20:52Engineering in quotes but I think you’ve already confined what you mean by platform so now you’ve also confined
20:58like what you’re trying to engineer um because then you’re like right well this is all about trying to drive
21:04Effectiveness on the right cloud in a consistent way without maybe the
21:10developers knowing hugely the big differences other than that they get to consume the things that they need for
21:15their applications that’s an outcome you’re striving for then yeah and without them necessarily being an expert
21:21in like a developer isn’t always an expert in Cloud infrastructure exactly exactly so that’s it so then you start
21:27to then you start to confine the decisions like what’s important so like well it’s not great if the devs are
21:33going to spend ages learning the cloud vendor it’s not going to be great if it’s a different experience from One
21:38Cloud to the next because obviously the speed of of onboarding from like one
21:43team that’s or moving the cloud is going to be slow versus another way they’ve learned it if it’s too too different
21:50then the technology is coming to play because then you’re like oh which Technologies are agnostic between the clouds is there any well yeah kind of
21:56Docker kubernetes containers are agnostic and that kind of starts to lend down your strategy then at those points
22:02but you also don’t want to you know as things we’ve spoken about before
22:07is when you do all these heavy infrastructure type plays they’re the
22:14things that are time consuming for porting between clouds applications easy change the library re-test your code
22:20probably way faster but on you know we need all this infrastructure then we
22:25need to get this team then there needs to be project manager then we need to do change management and we need to and you’re like yeah maybe this other cloud
22:32is like going to take his age like yeah because you’ve like basically wedded yourself in yeah exactly so like now an
22:37infrastructure problem not really an application problem and now you’ve blocked all the apps getting to the
22:42ground that they want to use right so it’s like so you have to take all those things into consideration it’s so funny
22:48that you say that right like if you if you think about why Cloud even exists it’s to solve that problem like yeah it
22:53used to take so much capital expenditure and time to go and build a data center
22:59right and now you’re like now you’ve invested so much in a single cloud
23:05provider and the infrastructure within that cloud provider that like that’s the
23:10almost the next bottleneck to solve to a certain degree so you’ve not like necessarily made it easier to Port those
23:18applications because it doesn’t really care too much about where the cloud you know where the infrastructure you’re using is
23:24um unless you’re you know you’re really um ridiculously um specific about the type of I don’t
23:32know CPUs that you’re using and you’ve optimized in certain ways or whatever right but for the most part you’re
23:37probably not um so that initial that upfront cost of
23:44time and money is really really high yeah um it potentially if you’ve made
23:50those those decisions and not enabled the right things in in the right way for the business because those things aren’t
23:56about the experience that’s a difference so if you’re to ask what’s the developer
24:02experience rafter well okay if business a or business Unit A has application a
24:10um an application B needs to talk to application a then the experience should be that maybe application B team can
24:16request connectivity to application and maybe that’s the experience we strive for and application either knows oh
24:21that’s someone’s requesting access from this other team I’ll approve or I’ll reject or go and speak to them yeah but
24:28that’s not that’s experience versus like oh actually we’re gonna have a prep we
24:33know we’ve already not even thinking about experience we’re just going in for like we need a network team yeah exactly and we’re going to manage it by change
24:39management and we do all this other stuff and it’s like okay well that’s very different now to like apps like
24:45optimizing for effective ways of working for delivering software optimizing for
24:50autonomy really automatic autonomy but also you’re trying to work out who should own
24:56the responsibility for those decisions and who should have the right oversight and there’s as I’ve said everything’s
25:02about trying to work out the right questions to ask on the experience ownership how’s the business set up then
25:10you drive the technology over it because you know what you’re aiming for um but it’s rare that’s what I’m saying
25:16it’s rare people thinking that way and generally speaking but um because it’s complicated to think all that through in
25:23advance and also just the industry doesn’t think in that way right the industry I think it does starting to now
25:30I think it does but I don’t think it’s it’s not an easy okay unless you’d
25:36unless you could Bridge again we’re talking about this before Tech and business yeah exactly then sometimes you
25:41don’t necessarily see the ramifications of how you’re working at scale because
25:46you’re Silo to a team you’re not necessarily seen that on aggregate this maybe isn’t the best way to operate
25:53right this isn’t the most effective way to deliver software um but you don’t know that because
25:58you’re kind of in your team you might be frustrated for other reasons in your team but you might not know everyone else has got the same frustrations in
26:04mass and actually they’re costing the business a lot or they’re slowing down delivery a lot and people will always
26:10want to go on the journey themselves and people think that this problem they’re seeing for the first time is unique and
26:16you’re like this isn’t unique a lot of the time it’s been solved and someone’s already experienced it yeah
26:22um so you don’t have to go through trying to work it all out it’s like just invest some time probably doing a bit of
26:28research to see like who else has found this problem already listen to a podcast listen
26:34to just our podcast Star podcast um yeah no I think so I’m kind of
26:40waffling a lot but um I think yeah for each time I’ve done that role historically I’ve learned
26:48different things anyway from the previous and they obviously feed into the next and then you start looking at
26:53things differently obviously you ask different questions a second time around the third time around the fourth time round
26:59and the industry’s moved so then you can have them to assess but I think I think there’s a level of maturity
27:04until some other radical kind of you know AI self writing self-generating
27:11World exactly which is when we’re like you know who knows what that looks like but I think for now there’s probably
27:18there’s enough if you if your assumptions are that the platform’s giving you enough value already
27:24um that they have become the platform because obviously they have a lot of products under their umbrella like a platform would
27:31um I just need to recognize that value how quick yeah and how quickly can I make that value be recognizable to the
27:38business units consumable and consumable yeah how quickly can I enable my
27:44business to consume the value that’s already offered by the platform exactly right yeah in a secure way yeah yeah
27:50exactly and I think then you have a customer as a platform team your customers and the other projects
27:55delivering and you have to care about making them as effective as possible with the decisions you’re making and
28:02take ownership of that and some of that is obviously researching the industry you’ve got to do that you’ve got to work
28:08out what’s already there um deciding what you’re optimizing for right because everything is a trade-off
28:13to a certain degree right so like you can um you could be uh or you can create
28:19this sense of security um by locking everything down that’s what we’ve seen in lots of businesses
28:26that’s how they used to function but um if you make it uh easy to do the right
28:34thing and um you know create sort of paved roads and all this kind of stuff which are naturally secure then you can
28:40also give autonomy to certain degree to those to those people so just understanding that there are like you
28:48have to be really conscious of the things that you’re optimizing for and there’s ways to do that well nowadays um
28:54and there’s easier ways to than having like a really kind of um
28:59uh really aggressive or strong opinion on on leaning one or the other that’s kind of why you look at a cloud platform
29:07and you start using you know picking all of the pieces to to build
29:14them up in the right way to to provide that internal platform yeah well I guess other things as well like
29:22companies that are more b2c have taken experience really serious right so it’s like I’m not an expert
29:30um you know on measuring heart rates or you know risk respiration amounts or
29:35whatever else right but my watch suddenly can do those things and it’s simplified to the point where I can tell
29:42you about those things because the app’s telling me right so that difference the
29:48same with your MacBook it’s like I don’t have to like it tasks there’s a new
29:53upgrade do you want to do it is is now a good time do you want to reschedule it like that’s about experience it’s like
29:58well yeah I mean I’m not I don’t have a local MacBook specialist in my area knocking door to door yeah you do saying
30:05excuse me but um do you need to do you need an upgrade for your for your Mac yeah exactly there’s a new thing come
30:12out we’re here to upgrade you and that’s sometimes how the industry is working and developers like you’ve got teams
30:17trying to manage infrastructure underneath of the business units that are really accountable and responsible
30:23for the outcome like I would be from a map but if I’m using it I’m using it right now not a good time I’ll come back
30:29tomorrow then and then maybe like that’s really inefficient how many of those people do we need but so you can see
30:34like actually can I sort that out yeah just shift it over and that’s where kind of shift left comes yeah you have you
30:41have shifted it left to the consumer yeah right you’re not shifted the choice and because experience almost a
30:49responsibility of that thing um being done well yeah but you’ve shifted the choice of when to do it yeah
30:55exactly when to do it and how simple it is and how simple it is right because I don’t have to get into the weeds of the
31:02upgrade yes right and then watching it and then things don’t work and I’ve got a load of things I need to do right
31:08um now it’s just tries to be as elegant as possible press a button you’re done exactly just it just happens or it
31:14happens when I choose it to happen um but I don’t get bogged down and all that so my phone I had to be an expert
31:21on literally Android itself for example you said I mean I’m not sure if this is
31:26a biased question but do you think that um people are trying to uh you know not
31:34necessarily thinking about the problem where they’re creating that sort of experience always
31:40um when they’re building an internal platform because they want to have this kind of Central
31:45um uh feeling of control I I’m going to upgrade you because I’m the one that’s
31:52responsible for this security risk and therefore you just have to comply versus
31:58no you’re the one responsible for the risk and you’re the one responsible for
32:03the relationship to your business users and this is going to affect them right so like where do you think
32:10um where do you think the in like that that internal platform team and how they’re thinking about it is thinking of those
32:17problems um well again I think that’s it isn’t it you need to decide what’s the right
32:25what’s the most well yeah what the experience results in probably some
32:32outcome because if that experience is good then maybe people can on board faster developers can onboard faster and
32:38maybe they can release quicker um maybe things are always upgraded right so actually if currently we’re
32:43saying on average only 30 of our Estates patched in time within say if our time
32:49was like 24 hours um or maybe it’s maximum five days how
32:55fast if we were to measure it is that happening across the business um because if it sends up only 20 30
33:02actually do do that and 70 or 80 maybe we are unpatched under the frequency of
33:09which we’d expect from a secure perspective then there’s an issue so then it’s like well how do we fix that
33:14issue well the issue is that there’s a bunch of people that have to go and work with the business units right and
33:21manually go off and do it um so we could scale the team one option right we need to find more of those
33:27people blah blah that’s going to be time consuming Etc so it’s just how you’re looking at the problem whether it’s
33:32worth solving if you care about it you’ll solve it if it’s made easy people will do it if it’s you know it’s just on
33:38how you’re going to design the this how you’re going to solve the problem and design for it in the right way if it’s
33:45not a problem then there isn’t anything to fix so in which case who cares right otherwise you’re creating a problem for
33:51yourself yeah um that maybe isn’t worth resolving as a priority but I think technology can like we’re saying
33:58technology can bend in a way that’s of value you can make anything work in any way you kind of want it to work within
34:04obviously reason physics obviously but um but I think generally speaking it’s
34:09about what it is you’re striving for and then also whether there’s things that
34:15are there I think is probably the point which you’re going on I think what I have seen is
34:21if people are on a journey themselves and then they can get caught in the Trap
34:26of um I was delivering for the project this is where this is kind of when you get um
34:33what’s the word you call it I don’t know my brain’s just kind of gone um
34:39describe it what you’re talking about oh it’s on the tip of my tongue it’s so annoying when it happens
34:45um the shadow I.T oh right right um so you’ll have a team that kind of sets up
34:52they’re solving a problem for a project then rumors happen like oh yeah like
34:59that team’s managed to get to like I think that team managed to get to Amazon even though she’s Amazon that team’s
35:04using Google through Google right so they’re like oh are they they’re using Google so they’re like hey we’d really
35:09like to use Google too and they’ve already done some work so like why don’t you just use what we’ve built like it’s
35:15gonna be so much quicker like then you can reuse but so you end up with this accidental
35:20platform accidental platform that gets created yeah just because it people see the value in something and then they do
35:26start to reuse it because there’s less friction yeah um and you’ve managed to get through the friction somehow that they’re facing so
35:33they’re obviously going to go to you exactly um so that’s like yeah so accidental
35:38platforms and how they grow and then having to kind of learn product management or maybe even not if that’s
35:45not a problem that you’re trying to solve wider for the business and then there’s like intentional platforms where
35:51you’ve got um people that have gone out to solve those problems that may or may
35:56not think about them in a way that um creates the right type of behavior for
36:01the developers and the right ownership for the people that are shipping code to
36:06customers yeah um so it’s it’s like there’s quite a lot of things to think about right there’s it’s it’s not an
36:13easy um thing to necessarily solve and it doesn’t really stop
36:22you know reflection I mean I kind of agree with monzo if I was to now you
36:27wouldn’t personally I mean it’s obviously Michael I wouldn’t build it was only because there was limitations
36:33to the existing Tech Stacks that I’d used before that I don’t think were optimal um and we’re quite frustrating and
36:39costly and time consuming and insecure all the other reasons and cloud back then was also maturing didn’t have half
36:46of the things now there’s loads of things there um it’s not finished by any means it’s not
36:52like you just it’s a it is a it’s not necessarily a developer platform no in that sense but it’s a platform which and
36:59you can create a developer platform from that platform yeah um so I think that’s probably the Delta but
37:05then there’s loads of things out there that maybe can get you pretty much most of the way there yep
37:11um because then you can spend time on the bigger problem like we all know there’s bigger problems like monitoring is it
37:17set up well do people have the transparency are you actually looking at um efficiency and latency and you know
37:25all those things in service meshes and they’re kind of like there’s so many other places to look I’ll probably solve
37:30problems but I think are you doing continuous delivery properly yeah can you have you put the right sort of
37:37quality measures and gates in to to enable something like that yeah this
37:42templating yeah exactly how do you get people up and running in really quickly what about the languages you know if
37:48you’re standardized on all that stuff do you have all like your logging formats worked out across all your languages so it’s all consistent do you have like
37:54libraries for observability in all your languages so it’s all consistent and so there’s so many other places that you
38:00can focus yeah um now but I think yeah it’s probably trying to work out where your Times best
38:07spent as a platform team to solve the other problems rather than going on Journey but I also do appreciate that
38:14you do accidentally go on Journeys without realizing and then you realize oh there was all these things out there
38:20I’m sure that happens to loads of people like I didn’t even look I wasn’t I wasn’t trying to build a platform so it’ll be that well it’s not intentional
38:28um so it’s accidental but I think when it’s intentional then yeah I wouldn’t understand why you like building your
38:33own car it’s like why would I build even if I was a mechanic I’m not going to build my own car unless it’s a hobby like why would I imagine if you you know
38:41if everything that you used you had to build it from scratch right imagine how much time and effort that would take you
38:47that’s nuts yeah I think if that was true I wouldn’t be here today houses
38:55don’t like frozen to death or something yeah I’m not necessarily the most practical
39:01um but yeah so yeah I mean what’s your view on uh on platform building building
39:08versus buying I think there’s there’s um my my slightly biased view is that
39:16um there’s so much upfront um kind of capital expenditure when you
39:22go on the Journey of building anything right you have to you know hire a team
39:28um put some time outside like understand the problem that you’re trying to solve hire a team in a very
39:36buoyant Market um so the salaries or day rates or whatever are going to be quite high and
39:44you’ve already now got this massive cost um and responsibility of of this ongoing
39:51cost um before you’ve actually delivered any business value and that’s again sort of
39:58the problem that cloud solves like you can just grow you you know all of
40:04that all of that Innovation exists behind the scenes yeah you can consume it in a way that allows you to grow with
40:11it and so I think that um going on the Journey of even like
40:18intentionally starting to build is just not efficient it’s just really really
40:23well you don’t have the economies of scale and you don’t have the economies of scale because you’re built it only for your own company exactly and it’s
40:30like oh yeah people are building it for lots of companies so they can do economies of scaling it so it’s like your price point internally as a team
40:37yeah it’s gonna be way higher because it’s absolutely just for you um rather than their product that’s for
40:43many and therefore you’re going to pay what probably like a much smaller price point yeah because of that you know be
40:49like running your own data center to match Amazons or Google exactly probably not cost efficient it probably can do it
40:56at a cheaper price point because they’re reselling it on yeah exactly so it’s just the same it’s just the same principle exact same principle yeah
41:02exact exact same principles and I guess the other thing is that like Innovation right so you know you go on this journey
41:11um you have this team you build it Build It Whatever and is highly skilled but then you’ve once you start getting users
41:19of that platform so you’ve you’ve gone on a journey maybe it’s taken you like a year um because that’s probably about
41:25the the average if you have a large sort of development team say it’s taking you
41:30a year to build this now you’ve then got a support cost and those people those really smart people
41:37that have built this thing are stuck supporting the thing and the users that they’ve built rather than carrying on
41:45that Innovation and that product that internal products growing with the actual demand of it properly
41:51um so but there is a lot I mean think about all the aspects monitoring logging
41:57continuous integration and deployment so much to do um like single sign-on tools artifact
42:04storage and then you kind of have things I mean some of those are in the cloud as well as Solutions
42:10um then you then have like well how are you going to package up your application how are you going to deploy that application what’s the Technologies
42:16you’re going to be using there and then how does all that feed into all those other tools in the right way how does
42:21app a talk to app B how do you see it goes on and on and on and I think those problems are
42:27complicated in some ways but like we’re saying they’re solved yeah really they’re not like they’re new they’ve
42:34always existed and there are a lot of solutions out there and lots of ways of making those things repeatable like if
42:41you you know and if you need monitoring lots of places the same ones there are repeatable ways to do it
42:47um but you don’t have to re-engineer it all from scratch and so I think that’s the
42:52thing so in conclusion use them don’t build them yeah unless you’ve got some very bespoke
42:59actual genuinely bespoke requirements which you might have I mean I don’t know but I think generally I say the majority
43:05I doubt very much that they were that unique and even if it was unique you
43:10don’t want to bend the 80 of the business to now be unique yeah for the small things that are
43:17unique because I do see that a lot like yeah but you know this project this has these requirements and there’s anything
43:22like okay yeah but like 90 of the other projects don’t so are we gonna now break
43:27all 90 percent a bit like really hybrid I kind of feel like Library Cloud it’s like you’ve got massive value in Cloud
43:34you’re now gonna what you’re gonna reduce the value of the cloud because you’ve got some on-prem exactly why
43:40exactly why would you do that I don’t understand this common denominator yeah let’s be the lowest common denominator
43:47and devalue the value of of cloud it’s like what I don’t understand treat them differently they’re not the same then
43:52exactly if it can’t match and they’re not like for light then they’re not like for like that’s it let’s not pretend they are and let’s not devalue something
43:59yeah where you can move really quickly yeah the cost you know the the price of
44:04moving and delivering is generally cheaper than doing it in in the kind of lowest common denominator yeah you’re
44:11not going to get the same services do you get do you get 300 services on your own product no okay then they’re not the
44:16same right let’s decide they’re not the same and maybe you don’t need all 300 which is fine too because it might be
44:22like actually maybe it does match you only need five I have um James as a service and Amy is a you know there’s
44:29all of these kind of people that exists as uh services so just you don’t know what to say
44:36it’s a good ticket a very different turn all of a sudden I don’t know what other business you’ve got
44:46technology here that’s what I thought was all right [Music] um brilliant yeah
44:52a wonderful note uh we probably uh um spoken about this effect yeah I don’t
44:57think we’ve just we’re probably just repeating ourselves um exactly if you’ve got other opinions obviously as well obviously always glad
45:03to I’m sure people from Spotify for my last episode to be like like what were you talking
45:10about all wrong yeah so if people have opinions to share obviously got to hear
45:15what other people think as well but yeah yeah exactly cool thanks for listening thanks for listening and see you next
45:21time yeah bye
Part of the Appvia family.