Revolutionizing DevOps with System Initiative | Episode #92

Revolutionizing DevOps with System Initiative | GTwGT Episode #92 - YouTube
https://www.youtube.com/watch?v=RfGLDvRsv70

Transcript:
(00:00) Adam Jacob the creator of Chef and one of the pioneers of infrastructures code is back with something new system initiative system initiative offers realtime simulations digital Twins and collaborative infrastructure management that goes beyond iac's limitations today we're talking with Adam to understand the technical challenges that led to this Evolution and how system initiative is changing the way Engineers approach Cloud infrastructure today we're talking with Adam to understand the technical challenges that led to this Evolution
(00:29) and how system initiative is changing the way Engineers approach Cloud infrastructure let's explore what's next for devops and don't forget to hit that subscribe alert and like button on YouTube also follow jtw JT on all podcast platforms of choice and now episode 92 Adam Jacob CEO and co-founder of system initiative hey Adam welcome to the show so just as a massive fan of Automation in general from my perspective it's great to have you on let's start about your background um cuz it's super interesting for people that obviously
(01:06) don't know but they should know I you're the found Chef which is just like guid likee status in this industry and thank you for all the work that you yeah and thank you for all the the work that you've done to you know get that devops message out there in Automation in general it's it's been a great ride to see that um tell us about your background though how did you start like you know go as far back as you want in terms of your own background and computers and how you got to kind of love automation yeah I mean that the tldr is that I've
(01:34) kind of always been a systems administrator I kind of I still kind of feel like a systems administrator even though I'm a CEO and was a CTO for a long time at Chef um and you know that started for me when I was a kid I found bulletin boards and I thought like running bulletin boards was like the coolest thing in the world which is which if you ever ran a bulletin board was essentially systems Administration you know like and I just fell in love with I fell in love with building the infrastructure that let other people
(01:57) talk on the internet and and that sort of what I've been doing the whole time and then I fell in love with automation because um you know I came of I I of grew up in the era where we were building isps and so that history included a lot of consolidation and so I started building automation to sort of consolidate different isps as as we came together so you people I worked for would buy another ISP and we would sort of slurp it up and make it a part of the same make it a part of our infrastructure and then turn off their
(02:24) infrastructure and the quicker you could do that the better off you were and then I had another sort of long stint with some folks who did a similar thing only they were buying applications and so we had to build sort of generic operating platforms that then we could take applications we weren't around for the writing of and understand them as quickly as we could and then Port them to some kind of management system and then figure out how to run them and so I just fell in love with that problem of how do you how do you automate large
(02:48) infrastructure and how do you automate things in a way that allows people to to use it and to and to sort of get more value yeah that's interesting I mean that's obviously the I mean I grew up in service provider land as well well bullet and board specifically I I consumed them at that early time of the internet I never I never hosted it but then as soon as I started working for an ISP ASP posting company getting into the back end stuff like you realize that you want to try and script things to make your life more efficient right that's
(03:15) what that's what it and Computing and Tech all about efficiency so if you can double down on that efficiency um it makes your life easier and that's it's a really interesting problem to try and solve yeah I just like the I like I like the metag game of it you know like when you think about how to build a system that like that search for like the right kind of primitive that allows you to like then solve more problems than you could have imagined because you kind of found the right shape of something like I really like that I like the chasing of
(03:45) that idea you know yeah and sort of and and then do you feel and you feel kind of good at the end of it the outcome because you hit a button or you run a script and it's like bam and you go well I did that but I didn't do that that's the whole premise of automation right like yeah and it's even better if you know if you build Frameworks that other people go use you know like Chef was so fun but the most incredible part of Chef was that like all these people and all these really ridiculous organizations brilliant people found
(04:14) chef and then used it to do things I couldn't have imagine doing and that was just like the most gratifying thing in the world you know like you'd and still today like I'll travel around and people be like oh thank you for building that or whatever and I'm like I you did it you know like I just I like handed you the pieces and then you made something incredible out of it and I just you know it was so fun to watch like yeah and we would't spend too long on shift but obviously you know given that background it's it's interesting to touch on it I
(04:39) mean how do how did you get to the point of I think I believe there was it was the application side of things where you were automating the application Acquisitions and whatnot that kind of led to Chef you know initially back in 2007 area yeah yeah we had a we had a consulting company um and so we we basically sold fully automated infrastructure to startups so you could like use pay us a flat fee and then we would automate everything so like monitoring and trending and application deployment and databases and backups and
(05:06) you know everything uh and the fastest we ever turned a customer on was 24 hours from like you showed us your app to we had a fully running fully automated infrastructure that ran your app in production and you know we were good at it um and you know the more we scaled both the the size of the customers that we served but also the number we served the harder it was and so that's kind of where Chef came from was we need a different a more flexible kind of technology that was also a little more reliable at scale um in
(05:35) order to actually serve our own clients and it turned out that once we built that that it was like pretty good you know and so we became a product company instead of a consulting company yeah so you were sort of eating your own dog food SL solving your own internal problems that's how that's how I think I believe that's how great companies start like I think about say um a few like like a few of the companies that you he started from they were devel game development company and they build an internal chat for the game and then that
(06:03) actually became big chat platform slack was the one I was trying to think in my head which one was I was thinking Tik Tok which one was it slack was that example right but interestingly enough like the people who were building slack were people who built Enterprise things historically you know like they built flicker first and then they built that's then they built slack and so like it's an interesting move from flicker to a video game but not such a big move from flicker to slack you know yeah that's correct yeah that's right is yeah
(06:28) there's something about building things that are the it's a lot easier to build great you know I I tend to think of product development as art and like it's a lot easier to build great art in the in this in the art that you know and love as opposed to being like you know if you're a painter and you decide to go be a sculptor like some there's some crossover between the two but like it's a lot easier if you stick to painting you know yeah that's really interesting in terms of where Innovation kind of drives from like if you believe in what
(06:54) you're doing and if you're working around that like I know some passion projects right so if you're building stuff from a point of passion you're going to put more into it you're going to be more proud of it and you're not just building it without soul I think a lot of um companies a lot of what we're seeing today specifically with you know all of the proliferation of startups and all these little tools that come out if you think about what's if you think about blockchain and what that built AI what that kind of created now a lot of
(07:18) those are built without soul I think the great companies and the great Innovations are built with soul I think it's impossible to build without Soul if you actually are trying to build something useful and you know in the same way that you can can't you can't make great music if it doesn't come from your own heart you know what I mean you can play other people's music and you can feel it and that's great and but you can tell the you know it's different when you watch like someone cover someone else versus watching the actual
(07:43) person play it when you when you can feel the the the intonation and the soul that comes out of it and products are the same there's a there's a there's a moment where you have to actually fall in love with what you've built and and it has to come from you in a way that's authentic and and you can't worry too much about how people will react or what they'll think because if you do then you tend to build things that are average you know like if you can ask someone hey what what do you think I should build what what do you think the answer is to
(08:08) that whatever the problem is and then you tell me and then I go build it for you and then I bring it back to you you very rarely look at that and go oh I love it that was amazing that blew my mind because it didn't blow your mind it was exactly what you said you know like it's not interesting it's not better it's not you didn't do anything to it you just you just heard what someone said and blatted it back out you know and like that that's usually a recipe for failure yeah I love that analogy so with that let's talk about let's pivot
(08:33) to what we really hear about which is system initiative and talk about that and you know where where did that come from like where where was the you know Chef was that you're doing the chef thing that kind of did its thing and then sh of initiative back in 2019 let's let's talk about how that came to be initially and the premise for what you're trying to solve with the company yeah so I mean with Chef I spent a lot of time in a lot of really large organizations and small ones too um you know figuring out how to help people
(09:00) sort of do that do devops and do configuration management and sort of think about automation more broadly and what I saw was that we we could tell people and there is a there's a path to putting the tools together that does provide really good outcomes you know like there are people who are great at this and they do really well but most of us don't the vast majority of us wind up you know like the door report now you can be a high performer in devops if you can deploy once a week with like a 40% failure rate I think or something like
(09:32) that which is insane if you think about where we started and what we were trying to do to be like ah you're doing great you can deploy once a week like that's bananas you know that is not what we set out to do and fail half the time and fail half the time and like but we do but we've moved the needle back because that's how that's how low the bar has gotten do you know what I mean like when you survey the world and you're like how's it going the answer is kind of mediocre and to me what I learned ort of toward the end of my time at Chef was
(10:01) that that that that that reason was not because the people who were implementing it weren't smart enough or because their problems were so complex and intractable they are complex and they are intractable but but that they couldn't do it it was instead that like the shape we asked you to work in you know the idea that what you would do is like build all your automation by writing code and then you would check it into Source control and that code would run and then you put it in a pipeline and we would run infrastructure kind of like we
(10:27) run applications like and then we would start adding in more and more stuff and we would shift everything to the left like that flow doesn't work very well and and when you put that flow into practice one of the ways you know it doesn't work very well and that it's a system level problem it's a design problem is that it still doesn't work very well regardless of the technology changes we put in place right like how many different technologies have we gone through you know Docker and kubernetes and chef and puppet and anible and like we can on and
(11:02) on and on how many CD platforms have you tried and and I'm not saying that that they're all interchangeable they're not they're all some a little bit better than the one before but the outcomes remain kind of sticky in the end that's interesting right because if it was a single tooling problem and it wasn't the total flow you'd imagine that you could improve it dramatically by finding the the one thing that sucks and and fixing it but you can't um and it's because that core primitive at the foundation of it was wrong and so that was sort of
(11:33) what I had come to believe um you know in 2018 and my co-founders also believed um and then the question was what do we do about it because it's one thing to look at the way that we work and be like uh how we're asking people to work is the problem and then you're like well but what would be better um because it's not like we came to work that way on accident you know what I mean we built we we we made a lot of choices based on really real feedback and really real experience and so you know we had to figure out how to do it and that's
(12:02) that's what system initiative is and that's how it started was that question of like well how would we do better if we were building a different experience if we were going to fun Al what would it be how does it not and how do you and and and why does it suck now like what about the existing technology can we keep and what about what about it can we do we have to get rid of like how deep does the rabbit hole go is an interesting you know was the interesting question yeah so if you were to give someone an elevator pitch on what system
(12:29) is of is at a very sort of high level what would that be yeah um I mean system initiative at the highest of levels is the future of how you're going to do devops automation um one notch deeper than that it's it's essentially a technology that allows us to build like digital twins of digital things so we basically model everything as data and then we stick it on this reactive hypergraph of functions so instead of writing code that describes what you want instead We Gather up all the information about the things that
(12:59) you have and we allow you to model them in really high fidelity and then we let you program that model program that simulation essentially by writing these reactive functions and then on top of that system we built a multiplayer user interface that lets us build all kinds of different ways to interact so if you think about things like Unity for example which is a similar kind of complexity if you think about the complexity to go build like Grand Theft Auto or the complexity to build like a big multiplayer MMO the number of
(13:29) different disciplines required to create that are roughly on par with the kinds of disciplines that are necessary to run a huge Enterprise and if you look at how they work together they do not work like we do you know like they do not push all of the all of the digital art designers to the left and be like right code to describe you know what I mean they give you a UI that you use for a particular job and then what you're doing is working on the data together to create the outcome the same thing happens in movies and all kinds of things
(14:00) and so essentially that's what system initiative does uh for not today for infrastructure but eventually for for kind of all of devops yeah so I mean what's interesting there is that you know you you you talked about what what would you call it the the graphical sort of element of or the work The Works Space that you guys have got where you do all the all the work and all the dragging and dropping is that yeah we call it like a living architecture diagram living architecture diagram yeah I mean that's sort of the interface that
(14:24) you use when you think about you know if if you have this big underlying this system is this really powerful hypergraph of functions programming that hypergraph of functions without some interface is annoying and so the that living architecture diagram is a way of saying hey I have these really high fidelity components that have a bunch of intelligence built into them how do I link them together and so you link them together by specifying these relationships between them which is really just passing data into each into
(14:54) each function right so you're essentially programming this big simulation by using this this living architecture diagram yeah you're talking simulation digital twins like multiplayer gaming it's it's a fascinating concept the thing that I go to straight away when we think about devops world and what the mindset of those people are is that they're so anti UI so anti- interface it's all bit about code it's all about getting rid of that because that's the old way of doing it yeah and now and you're kind of going you know
(15:27) what let's let's let's let's go back to that's and let's use interface to kind of through it the hyper situation which is interesting yeah I mean look I'm as surprised as anyone so you know like if you if you were if you were if you were hanging out with me when we were talking about whether this was what we should do I was like this will not work this is a dumb idea and it turned out that it did work and it wasn't dumb and and the reason is because look most of the time in our space we have a good reason to
(15:56) have an aversion to these sorts of tools they tend to be toys right historically that's that's what they are like they they they don't scale in complexity they they aren't useful over a long period of time they don't work well with multiple people you can't keep track of them you can't use them in the Enterprise you don't get audit logs like there's a million reasons why those tools haven't worked in the past it turns out that that's also because those tools lacked a fundamental primitive that allowed you to actually think about modeling The
(16:25) Domain in a way that makes any sense at all and so you know if if you're trying to build a visual tool that sits on top of arbitrary source code it turns out that doesn't work very well right people forgot like 4gl and some of those things from back in the day where you know there was a minute in the industry where everybody was convinced we were going to use uml and thenl was going to like generate the code for us automatically from the diagrams it's going to be amazing it didn't work very well because you wanted to customize the code all the
(16:51) time right and so we wound up not doing it because it turned out not to be flexible enough like that's what the history of infrastructure tools are in this space too right it's if they start from the position of Simplicity if it's like I'm going to simplify this complex problem well then they don't turn out to be powerful enough and infrastructure is a domain that requires essentially infinite power and so part of what makes system initiative great is that it doesn't give up any of that power like because of the nature of the domain the
(17:21) nature of infrastructure allows us to build these digital Twins and to model things in this way and then we put the code back into it by saying what you program is the simulation and that's the thing that's a bit of a head trip because you know when you think about it you're like okay you know if you shake it off a little it kind of starts to make sense um but it's a big it's a big jump yeah the concept of toys versus tooling is really interesting I think about I I I would get super excited if I was I had a new
(17:50) toy which was a tool to try and do something better in my daily line of work right but we would the amount of the amount of things you would sign up up for because this was going to be the next thing and then that is a toy right the by definition in our world because we want to play with it we want to see what it does have fun with it we want to see if it makes us feel good so I get that analogy it's really interesting um talk about the living architecture and digital twin component of it because I find that it's all about the modeling
(18:19) right so talk a little bit about that and how that is so fundamentally like you said like shifting for the mind yeah so think about it this way so you know when you bu an application right so we sit down we write some application code like what tends to happen with it is it gets like compiled into an artifact the artifact sometimes gets deployed it gets it gets put on things it gets put in container it goes and run somewhere but like we kind of know what the flow is that that we're going to do here right um and that that flow is pretty sticky
(18:49) kind of regardless of the application that you're writing infrastructure on the other hand like is not like infrastructure things happen to infrastructure that you don't want to have happen right data catch on it's physical things they catch on fire they they um they they disappear people will change them like all kinds of State happens to infrastructure because it tends to represent physical things even in the cloud right even even the digital parts are not they're not as they don't work like you think that they do and so
(19:19) part of what we do is we took you know we took lessons from the real world again and you know how do people deal with for example a really difficult ult problem in our space is is feedback loops think about how long it takes to know if you have a lot of terraform let's say you have a a reasonably large terraform repository how long do it take to run a plan and how long does it take to like run the apply those numbers can get up high really fast right tens of minutes 20 minutes right and the feedback loop to know if you've if
(19:48) you've done something right or wrong it's it's a little like driving a submarine you know like you can you can tell but it's a little tough and so I'm laughing because I get I'm laughing because I get this I'm laughing because I get this I know I know because you're there yeah yeah so think about it like how do people solve this in the other in the outside world so like let's use Formula 1 cars as an example so you know Formula 1 cars very complicated incredibly difficult things and they get a very limited amount of testing right
(20:13) you can only you can only put them on track a very small amount of time and so what they do to solve this problem is is they build a simulator they build a really high fidelity simulator of the car and they run the car over and over and over again in this simulator with all the different setups and and mechanisms they could possibly try so that by the time they get to the track they only try the ones that are most promising right and what they're doing is building a digital twin of the car right um we see the same thing with like
(20:39) miners right we we tend to send robots now to the bottom of deep mines we don't send people um and those robots there's a digital twin of the robot up top where somebody is controlling all the different pieces and all the different inputs and you know we're combining the sensors in the real world with the feedback loop at the station where they're controlling it and like and so we do the same thing only for infrastructure right so we build these really high fidelity models that are one toone to the Upstream so you know if
(21:07) AWS describes their API in a particular way we model that API one to one right we're not building like an abstraction or a declaration and then we program that with functions right so you you construct that High Fidelity model in this in this way um the advantage of doing that is that we can track both sides right I can track what you say you want and I can track what it actually is in the real world and I can do them separately and then I can visualize both of those things for you so I can say hey the real world changed and you can be
(21:40) like yeah that's right because there was an outage and I just went to the AWS console and fixed it because I wasn't sure how to do it in terraform uh and so in system initiative that's no big deal you would just see that there's a Delta and you would update the component and then that'd be it you just because you don't really care where the state came from what you care about is that it's managed right you care that it's been understood you care that you can see the Delta and the drift yeah and I think from a visualization point of view
(22:06) people should probably go and check out the website to get an understanding of you know you've got some great demos on there that shows the interface and what what you mean by you know things not being right you've got indicators as black and white as as red and green right in terms of what's going on is something figured um you drop elements into into into the main sort of flow there the elements are AWS at the we'll talk about you know why what's going through so ec2 instance you know you got a VPC you you've got a gate API Gateway
(22:34) and you're kind of connecting them yourself um In This Very sort of real sort of way much like it almost reminds me of like you know the old database table days we trying to link databases and keys and whatnot that's kind of how it looks like to me anyway depending on how old you are it has a Vibe like depending on how old you are it has a Vibe kind of like uh almost like Visual Basic um or sometimes it almost gives you a Vibe like small talk if you're really old but like yeah it's uh it's it's it's like all of those things were
(23:04) inspirational in thinking about what was possible and it turned out to be a really good fit and like I said I'm as surprised as anyone because yeah you know like I made my living building declarative uh declarative code and telling everyone that that was the way and look I'm proud of what we did we weren't wrong like the the current you know the state of the status quo is much better than it was 15 years ago so I'm proud of it but we can do an order of magnitude better yeah and I think to be fair like you also expose that element
(23:32) still within it right like people can see that the Json configuration they can see the actual code that's running so it's not like you're abstracting that completely and hiding it which I think it fits for both both worlds right kind of making people happy on both ends of it in many ways we're actually exposing more of the code so like one of the things that happens here is like if you think about how like terraform or Chef or puppet or kubernetes is built like you have this interface that you can see so so you know kubernetes you have the
(23:59) yaml but then there's all the code that runs the crd that actually reconciles that thing and manages its life cycle um in system initiative those things are reactive functions that are also part of the graph that you can change in real time so like you can just open up the way that a particular if you like for example we had a a user who uh needed to use the Mac features of ec2 and we hadn't quite and we hadn't modeled it correctly and so he fixed it he just opened up the assets we shipped by clicking a button which opened up all
(24:33) the source code and then changed it and then there's a little button you push that contributes it back and he hit that button and we like took it in and vetted it and then we shipped it to everyone now we have mac support and he didn't have to like open he didn't have to go Fork a provider in GitHub to do that he just looked at the source code that was running in his local graph changed it saw that it worked and then said now I would like to share this with everyone else and we did it and like that's because what you're doing there's plenty
(25:02) of code but what the code is doing is programming how the simulator works really cool yeah so talk about the multiplayer aspect of it and the g the slight gamification is it a gamification element or is it more multiplayer in a different sense I mean it is kind of gamification it wasn't we didn't I didn't necessarily think of it as gamification but um yeah you know if you've used notion or you used figma or you've used Google Docs then you know what we mean when we say multiplayer collaboration like it it basically it
(25:29) happens in real time so if you and I were to log into the same system initiative workspace right now and and we were in the same change set which is basically like a branch so but it's a branch of everything both the source code and all the properties and it's where you can you can change the simulation safely basically without having any impact in the real world um then what would happen is we would see each other's changes happen in real time like you'd see my cursor you'd see me drop things in the diagram I'd see the
(25:57) things that you had appear and could change each other so like it really changes things like review you know if you think about code review right now you like you like write a bunch of code and send someone a diff in system initiative you invite them into a change set and they can like look at everything you did and they can see it and if they want to change it they can just make changes and then when you come back you can see what they did and all those things are are represented to you in this diagram and you can have a full
(26:21) audit history of everything you've done but like the big transition for me like I had forgotten what it felt like to do that kind of work in real time with other people you know like the last time I remember doing that kind of work that way were was like building an ISP with one of my best friends and sitting around building system initiative with system initiative with you know watching the our team do it I was like I was emotional about it you know I was a little moved because I forgot how much I wanted to do that like I forgot how much
(26:54) this was actually like it was actually it it's always kind of been a thing that you do with other people because no single person really has enough knowledge to pull it off and that was kind of the core thing that devops realized from the beginning and only we lost it you know kind of in code review and pipelines and Incredibly long feedback loops and and like now it's back and like I honestly I feel a little I feel a little emotional about it because I'm like I missed it you know I miss those people and I you know I think
(27:25) about I've been eight years out of a data center right and now but spent first portion of my career always in data centers building you know building getting the getting the smell of that new that new T and putting it together and then installing and then but you're right it was never a single thing whenever you did that whenever you deployed a new DC whenever you got new kit it was it was a group effort team effort and it felt good doing that I'm really I'm really like you've tapped into something I I didn't know I missed
(27:54) it I didn't know you know like and but I did and and and now that I have it back again in this different very different way it's not the same as being in a data center but but it is the same as like as interacting with other human beings again and and when you think about it in terms of expertise like another good example was somebody popped in our Discord today and uh they were trying to figure out how to use Lambda and they never used the details of Lambda they'd only like sort of used the high level things and so they were trying to figure
(28:23) out like how the IM am rules would work for the for the role for the Lambda they needed to use and one of the things we offered to do was like if you want us to we can just hop into your workspace and like one of the folks on our team who understands Lambda really well could just show you what to do and like literally do it and then you could like watch it pop into the change set together and see if it was good that's wild you know it's crazy how much better it is that is C that's that's that's getting back to that feel like you said
(28:49) I I really enjoy that point talk about the 200% Pro problem because that's something that is you know I found interesting as well yeah so I first understood this problem with Chef so you know when we first buil chef was configuration management right and so the thing everybody used when we built Chef was Apache 2 do you remember Apache 2 the web server um or engine X if if you're if you're if you're not cool enough to remember aache but I Pat I Pat you all the way I mean I've been using engine X for a while but anyway the pro so so the
(29:22) thesis here was always that you would have a single incredible cookbook or module or whatever and what it would do is cover all the functionality of the underlying thing that it automated so nobody would ever have to write it ever again we were like one person will just automate enginex one personal automate vpcs you know and then nobody ever has to build a VPC by hand ever again they just they get it done correctly because you're driving this higher level abstraction and it turned out that was dumb because what happens is you have to
(29:50) to use it you have to know 100% of the capabilities of the thing you want to automate and 100% of the abstraction that whoever automated it for you you put on top of it in order to let you do whatever it is that the thing you understand can do and in fact you in any given moment probably need 5% of all of the possible things you could configure and you know exactly what they are you're like hey I know how to write the engine X config now the question is how do I map it to this crazy thing you gave me and so if what you gave me and you
(30:19) see this in like a terraform module if you ever grabbed a terraform module that has like or a Helm chart that has like 900 variables in it yeah and you're like oh my God which variable do I what what do I do and what you're doing is then mapping those variables back to the documentation of the thing that it's automating for you and so like did it actually make it better for you or make it worse and what we learned in the chef Community was it was worse that actually if you just built the your own cookbook to manage engine X that only did exactly
(30:48) what your application needs it was easier for you to write it was easier for you to maintain it was easier for people to understand like it was better and so we just stopped trying to have like big Community cookbooks that just did everything for you because it was worse um in fact little less was more um and what happened in system initiative was we discovered that that 200% problem is inherent in that that moment of abstraction that that thing where we map how the underlying system works to like the declarative Paradigm is really the
(31:22) source of that 200% problem because I have to map every option to this to a declaration and it's not usually declarative and even when it is it's a little weirdly sticky and so you kind of you kind of need to just model the system as it is uh to avoid the problem yeah yeah yeah yeah it's almost like the premise of trying to create less work through efficiency actually creates more work yeah it happens in particular when you're trying to do it at scale when you're saying this substraction will replace this thing with equal amounts of
(31:55) power which is usually what happens and so that's why you can and it happens you can see it in every module store right like go look at a Helm chart that automates postgress and it'll have 9,000 tunables because postgress has 9,000 tunables and some and everybody needs one of them you know and then you know and it just it it becomes combinatorically weird and system initiative avoids this problem by saying let's not abstract them at all let's just model them one to one and then let higher level semantics Drive their
(32:23) behavior so you know imagine having a component that configures postgress but not because it abstracts postgress but because it understands postgress and it knows how to tweak the variables for you and you can watch it happen and so you know the a similar example here is like let's say you have a template uh in most things where you know you you fill out a template and then it goes and does some stuff for you you can't usually then edit the results of the template right not if you want to keep using the template because if you change it then
(32:55) now suddenly I have to grow the template and I have to do all that stuff stuff that's the 200% problem in action whereas like if you don't if you can just model everything one to one and then let higher level semantics Drive their behavior which is pretty natural to do in a graph with reactive functions right because the answer is just okay well that property is set because this other property over here sets it and then whenever this one changes we should react by by by changing the the value interesting hey quickly not a lot of
(33:24) time left but let's talk about the open source aspect of this um on gith um you know you talked about a semi Community contribution with the Mac thing I guess that's kind of an example of that but maybe more of a feature of what you guys are done there but open source Pro became problematic this year because of what happened with terraform I think and you know that sort of thing and people started to question the legality of it we had the WordPress thing a couple of weeks ago as well that kind of caused issues you know what's
(33:51) your general take on this and why have you decided to kind of go down this this this path well look uh there's two there's two very different parts of me in open source and we could spend 40 minutes easily talking just it's hard for me to do in two but I'm going to do my very best but look the the the short version from a business point of view is this is transformational technology and so today a very small number of people in the world are going to really understand what it is and understand the implications and you know I'm on this
(34:19) podcast because I'm hunting for those people you know I'm hunting for that 2 and a half% who are going to hear what I say and they're be like oh yes what if it was a reactor graph of functions and what if we did think about it as a data transformation across it and what I can't do is Cap those people's upside I can't also say to them hey I have this transformational technology for you but you can only have it if I'm the only one who gets to benefit exactly the way that I think I should instead I have to say
(34:47) look this is going to change your life I don't know how I don't know I don't know what you're going to do with it but if but if but if you if you can use it to change your life the way I'm using it to change mine then we're both going to be better for it and we're both going to thrive and my business is going to be bigger and whatever you decide to do with it'll be better and if we can figure out how to do that together as a community there's no question we can do more together than we could do Alone um and you know it's more it's more
(35:14) complicated for my business because you know I have to I have to think through a lot of those Edge conditions about well how am I going to relate if somebody does something I don't want them to do and the answer is I'm just going to sell them to Fork it you know like there's I have to be comfortable with lot of those decisions but I'm very comfortable with them because I believe that that we can be a good Steward of that Upstream in a way that allows everyone to thrive because because it is transformational
(35:38) technology and that's the best way to bring transformational technology into the industry if you don't open source it it won't it won't live up to its potential yeah and I think that ties back to what you talked about developing and creating stuff with soul and purpose as well I tie those two things together what you just talked about right that joury so that's that's really good exactly I didn't I didn't I didn't design it to not change you I like I designed it to to like I what I want it to do what it's built to do is is change
(36:07) how this whole industry functions at Like A Primitive technical level and if that's what it's going to do it has to be open source awesome he just little bit of Futures we've got like two or three minutes left so obviously at the moment AWS is is is is the is the focus obviously I can see why that is huge huge sort of addressable Market where do you see it going like what's what's the plans for just like in the near term and then where do you want this to be in sort of 12 to 18 months yeah so I mean a lot of what we're doing right now is is
(36:36) adding more is adding sort of broader capabilities so you know things like really deep like rebok systems for Access Control and and policy adding more asset coverage like you said um adding in like really comprehensive audit logs they're already there but sort of exposing them in a better way um you know doing having more ways to sort of cope with the visual complexity you know it's really what we've done with the living architecture diagram is really cool but it's a bit like staring at the sun right now because it's like a
(37:05) huge Infinite Canvas which is on the one hand amazing because you've probably never seen what your entire SAS environment looks like your whole production environment looks like all at once but also when you can see it all at once it's kind of overwhelming so like some of that's going to happen the big thing that's going to happen in the next 12 to 18 months is that we're going to learn together with the people who use the software exactly what they need to be most successful and then it's going to really go nuts you know like as soon
(37:31) as as as soon as what's happening is we're learning in collaboration with real people with real problems and we're like figuring out how to overcome those things using this primitive together that's how that's how it's going to really grow um and and turn into what it can be so I'm really looking forward to that Loop because that that that symbiotic Loop between okay here's this really great primitive we've built now people go use it then they come back and tell us how it was and then we react to that and then we talk to them some more
(38:01) and then we build some more and then they react and we build and you know you just sort of spider your way to Greatness yeah do you say it just very quickly do you say it extending out to say on premises sort of environments workloads maybe even desktop like a local so if I've got my Mac menu I've got a little Dev environment here doing its thing like you know do you see it kind of extending there you can model anything so so yeah I think I think it's G to go where people want it to go you know and like VMware for example I think I think
(38:32) system initiative could very quickly be a better interface to VMware than VMware by a lot like if you think about that living architecture diagram and the things you have to manage in VMware like let's go you know that's that's that's needed yeah V we need a boost as well just so you know you know what can go they could use a hug they could use a hug for sure and so so you know like but I I think so yeah I think I think all of that's coming and and what we have to what we have to do together as a community of people is
(39:03) like we have to learn how to do that we have to learn like well where are the sweetest spots because it is a new technology you know and and any new technology you got to figure out like okay what is it and how does it work and where does it fit how can I play and you got to play with it and make it go so awesome well I would encourage people to get onto the system initiative website have a look at the videos actually go in and play you've got a free tier as well I think from a certain level so that's really great so I've I've even G played
(39:28) with it already so I'm just getting my head around it but yeah this has been an amazing look I really love what you've done here um and I think more people should go check out system initiative so thank you very much for being on episode 92 of great things with great Tech 92 nice work getting to the2 nearly there yeah incredible thank you for having me no worries hey just as a reminder thanks for listening to this episode stay tuned for more episodes where we continue to highlight companies and Technology shaping our world don't
(39:59) forget to follow us on social media jtjt podcast and visit jtw j.com for more great content and all past episodes if you enjoyed this episode make sure to subscribe on your favorite podcast platform and on YouTube please spread the word and if you feel like it drop a review thanks for joining us and we'll see you next time on great things with great take [Music]