Codiac: Kubernetes Doesn't Need To Be That Complex

Key Takeaways

[Music] you're watching the new sack makers a podcast for people who develop deploy and manage at scale software for more information and articles about at scale Technologies please visit the new stack. now enjoy the show Kodiak streamlines the software development process by eliminating traditional devops responsibilities for engineers and redefining their approach to software development product management and cloud architectures with Kodiak engineer can fully integrate and optimize the sdlc process all right everyone today we're learning about Kodiak the platform why it's needed what's changing and how to think about it in terms of the software development life cycle I'm here with Mike fral who CEO and co-founder of Kodak hey Mark how you doing Mark is on the road what what what what highway you on what wristop what town High 10 and I'm just outside Lafayette Louisiana making the trip between Houston and Tampa oh right on and Ben gazi co-founder of Kodiak hey Ben hey how's it going where you where you from where you where you uh calling him from today I'm in Houston right okay yeah he's he's leaving he's leaving my part of town going back to his place okay good well so we're going to talk about the Kodiak platform we're going to talk about why it's needed and what's changing again and so what we want to do is just give a general overview here a general overview of Kodiak and then we're going to get a quick demo and then we'll finish finish it up with some more conversation so I just wanted to start you know by asking about the core technology itself we're coming up to cubec con and there's a lot of people are going to be looking at their platforms and and how they work you know what are the what are the company's roots and what problems did you see that set you on this journey that you're on Mark sure uh The Roots basically we've been application Architects uh for the last couple decades um and watching what was going on in software um and and and devops and whatnot it was started to get to a point where hey look there's a lot of great out there but we haven't really productized all of it and and there's a lot of good opportunity to say hey the stuff that today is no longer art but science is stuff that I want to get quick easy out of my way so I can get onto the next piece of tech and start thinking at the next level and and that's really kind of was really the birth of Kodiak to say hey look there's this opportunity um when the IAC was out there the cloud is out there um um and and kubernetes walks into the door we have orchestration we have all this all this opportunity to put all these pieces together and make a really great Hole uh that really sort of propels the the product development and basically we Ben and I both put that uh put that idea into practice and uh and went for it couple about five years ago Ben you must have some stories about those early days working with kubernetes yeah we I have stories ofet today yeah it we've got some wild stories um especially the early days of of Kodiak we sort of always carried a couple customers with us as we built the product and uh you know Consulting on this side build a product at night right kind of thing and yeah I mean there's kuet is is is like the best place in the world to have your app running and like the worst place in the world for a team to run an app it's like it's it's wildly uh difficult and and esoteric and Nuance right and it's like yeah but once you get it working man it's perfect it's beautiful fantastic um so yeah we as we were building the product you know one of our biggest things that we bring to the table is is ingressing and we can do it programmatically we can do it automatically and we do it really well really well with C uh I can tell you from those really really early days the first year we were building this thing and we just could not get the thing to work properly work reliably right and then finally we had that aha moment right with ingressing where it's like that changed everything it's so good um and now our customers love it like that's that's one of our when I do these demos like the one big thing is well there's a bunch of different things but everybody always talks about you know you we even got some really really fun terraformed guys that they do this stuff for a living all day every day like you can't do that you literally can't do that and we're like look I don't know what to tell you it's already been done like it you just you just did it it like the system did it for you it's done so these aha moments are are pretty fun to watch people you know get it well let's get into the demo maybe you could talk about that you know that Ingress capability that really change things and show us uh you know what this makes uh customers liking it so much yeah absolutely um so so Ingress is sort of um a piece of it and Mark and I really wanted to take a a holistic approach yes great you know automation um so yeah we'll start there let's do that um all right so here we are inside of Kodiak and when you when you jump in the first thing you see is all your environments on on one screen together right so one of the biggest problems in the industry is I just don't even know where my stuff is like I don't even know where in AWS do I go to find this where in Azure do I go to find this like how do I put all this stuff together so you know the first thing that you see here is like one pretty clean screen that's that is basically like a set of environments with a set of um boxes in them these boxes we call them cabinets they're they're tied to Nam spaces but there there's a little bit more going on than just namespaces here um and so you have a multicloud multi cluster implementation of your software company running here on one screen so already we're already sort of sort of level setting like what your experience is going to be here um and so we wanted to be able to say well look at any moment I want to be able to you know deploy that thing out or manage this container or you know set those configurations um but I don't want to have to know all the esoteric syntax of of how to put all this stuff together so if you know take the dev environment for example example I've got a Dev environment and it's running you know um my systems it's running my products I've got a whizbang product in here that's got an API on it I've got a front end another you know public API it's all running different versions Kodak kelpy version this stuff um but also will tell you like hey you know this is running out in this cluster uh Ben created it here are the things we're talking about with this version of the system um here's all of the history of the dev environment you know successes and failures right who did what when here's all your microservices that we deployed um and so we were talking about ingressing right and so we can now have a conversation about um this one doesn't have any but I'll go to one that does have Ingress uh oh doesn't there it's a demo we're always going to have demo issues with demos it's quite the clean interface it's uh it's really well well done yeah thank you so um let's just get let's get a new environment right so what Kodiak what Kodiak can do and bring to to the table you know everybody's asking what is Kodiak it's infrastructure on demand it's container in in cluster management and it's software development uh life cycle work so when a team member says hey I need I need to work on that payment Gateway um they can literally come in themselves and say I need a I need an environment for that payment fix right and I also need to be able to say I know Mark was working on that payment thing I don't know the version number I never remember the version number um but it was it was in that Sprint uh 14 uh where is my stuff did I delete all my tags here you know while you're looking that that's a real commentary on gitops isn't it I mean it provides you you know I know that you do provide snapshots and such but being able to look back at that history very simply through an interface is um you know looks like it's a kind of core aspect of what you're providing that's exactly right I mean that this is we want to be able to say look the Nuance of cloud te is is insulated from you or for you and now you're able to just say well look I I am looking for that payment Gateway thing um let's dump that into my new environment oh it's a drag I also need like I need my front end in there so let's go find my front end and we can put this oh here's the portal fix thing yeah so whoever tagged this is Portal fix that's what I was looking for and so that's great I also need that that rdus worker to get my payment thing working so I have tags and all these things so bang I I can put together an environment that is configured memory and CPU all the infrastructure is done is completed for you we'll even pull secrets and set secrets in the cluster for you but we also do ingressing right and so we can we can now say well great this this thing is up and running and and healthy so you know we can go open this thing up takes a little while for the searchs to CT up but once they do we'll have this environment working for us and we'll be able to go use it and work on it and and have you know have a team use this environment lock stck and Barrel right so now we're we're shifting that conversation from you know uh do I send it through a cic pipeline how many microservices do I need to get in there who's configuring these things who's managing this stuff like now you can just actually go do experimentation and play with this stuff and work on it yourself and and it's all done through the API correct 100% of Kodiak is apid driven right so we have we we API into kubernetes we API against Azure we API against AWS um in fact this this front end here is a front end to our API which is running on my local machine so when you start Kodiak you're actually starting a a running container on your laptop or inside of cicd so it's sort of this unique on prish type of solution right you know yeah we're we're basically managing all your a AWS and Azure tokens for you so that you don't ever have to relog in right so like remembering your adws for East US remembering the token for that you know keeping keeping track of this stuff is that alone is mind-numbing uh when it comes to like being productive yeah Kodak just does it for you we just we manage all those tokens for you we manage these these multicloud implementations for you but again Mark that speaks to how kubernetes is great for managing applications but the people part is a real is is a real trouble Troublemaker for people absolutely yeah the the promise of kubernetes I mean since the first day I walked in the door years ago is phenomenal it's it's that moment you know I remember the first time I used it and I saw what it could do um it was sort of a culmination of all these things that we've been trying to do for a really long time and and here it was and then you find out what a steep hill it is to climb to to get all those pieces together and then on top of that you you know having having to perform a small task is not that big a deal having to uh write a small script or uh you know Mark an update into a into a yaml file somewhere that's not that big a deal but having to do that for everything and having to do that for say I've got 30 services that are running all these different versions I've got all these different environments and in fact I want to create a new environment on the Fly how do you do that by hand it's it's just not going to happen and and so you know the ability to say I can collaborate I can collaborate on on on what my intuition is which is I need this version of that thing over here so that I can work on it and then we can start having a better conversation with the guys we collaborating with and the women that we're working with on the uh on the other side so like anyway that the the facility of doing these very basic things but doing them on mass um really helps the team develop a better workflow and and a better uh automation strategy so I I wonder about how your platform manages the actual scale that comes for a lot of organizations and I think we're reaching that maturity G curve with kubernetes where people are realizing wait a second I don't need a cluster for everything I you know you know I should be using kubernetes for what it's good for and that's for working at scale but the scale becomes quite complex and that reflects changes that you need to think about cicd it reflects changes you need to think about you know deployment and efficiency in your deployments how does your platform manage that is all being done through the API and do the users get some ability to kind of control that you know through their through their interface yeah uh just like uh you know all the Kodiak features um there's a everything is designed for the intuition and everything's designed for reuse and repeatability so it it may be simple like I was saying to write down hey let's make an HPA a horizontal pot autoscaler um that works off of memory and another one that works off of CPU and has these performance characteristics um you can do that in a short period of time um it would be nice to be able to copy that to another asset it would also uh be nice to be able to say well I'm going to do that but I'm going to need that done every time we deploy and and do I now have to script that thing out do I have to what do I have to do to that to make it repeatable and also if I'm the only one that can run that script or really knows how that thing works or really knows how that was put together that's just not very good for the team what you really need is you need to be able to say hey I put an HPA out there anybody can use it anywhere in our our ecosystem just hit go um that's a really important part of being able to move fast um and and make changes so yeah I would say that there's there's um there's a big component here that I think we're we're walking around and it's it's this ability to you know your point about scaling out uh your Ops your Automation and your coges is is really a you know it's an acute issue for everybody that's doing it and the the real problem is State Management you know knowing where my stuff is and how it how it's supposed to be put together so Kodiak brings that to the table we're this Central orchestration for all of your clusters across all clouds right so when it's time to Sunset that old 1.22 version of of kubernetes you can literally take all of your objects out of that cluster put them on a new cluster with a dragon drop and you're Off to the Races to your point you were talking about snapshots yeah that's what snapshots can do for you so I'll show you what that looks like that's that's what I was curious about yeah great love to see that yeah while he's putting that up uh real quick uh we've had plenty of customers come to us talking about kubernetes scale saying um you know that's what I we got into kubernetes for it's just that we haven't gotten to it yet because we're really busy and we don't have the time to put into it um that's where a tool like this can really come in handy Mark with so this is I'm going to add a new environment uh mark this uh mcf a West to that's a good one for us to play with huh it might be that might be an old one these are well that's great let's see if an old one will go um these are all clusters right that that Kodiak has created for us and okay creating creating a cluster through Kodiak is literally like seven seven you know questions in our CLI um anyways let's let's go have some fun with it let's see um I want to create a new environment I want to call this uh Alpha and I want to put it on this old or this cluster instead of this because this is where my production cluster is now um so let's put it on this new cluster and here's what snapshots does for you so that whole Paradigm of you know broke up the monolith now we're in containers and microservices container management is a utter nightmare nightmare and and honestly it shouldn't be like containers is beautiful and it's awesome but managing a broken up system you know five apis over here a couple of front ends couple workers over here like it's it's incredibly difficult right so like Kodiak is saying look yes container management is absolutely crucial you know on a component level but there's something higher than that there's there's the system that we need to be thinking about and worried about and that system is what nobody is is really focusing on and so we're saying well look every time you deploy a a workload a container to any cluster that you that you've ever created with Kodiak every time you do that deployment if we haven't seen this collection of of workloads together at these exact versions we're going to say hey that's a new system version and that's a snapshot so now I can say I have a perfect snapshot of what is happening in my production blue environment I can put tags on it I can have metadata on it um I know for sure if I want to recreate my production environment I can pick this thing up literally and drag this off into my Alpha uh environment so it's going to pick up all those microservices that were in blue at exactly the right versions right and now I get to go literally you know deploy that to Alpha uh get this new environment up and running and healthy and simply switch my Ingress over from the old cluster old production to my new cluster my new production and now we're cluster hopping and leaving the old clusters behind right ah so that's that's snapshots it's it's system management right it's not container management it's managing the collection of containers can you then create ephemeral clusters then can you do that ah Mark do you want to take that question uh ephemeral clusters yes I mean we you create your clusters in Kodiak and cod we do that with the CLI and you can customize obviously choose whether you're going you know which cloud provider you're going to put it on and based on that cloud provider what specs you may need um that cluster goes up and then you attach it to an environment and so these these uh clusters that you see down here are the ones that we have attached to that environment and so now your clusters are uh they're now sort of in that in the in the system ready to be used for by a cabinet that that Ben's showing you right here so again uh I add a new cluster I attach it to an environment I'm done with this other cluster I can drop the cabinets that are on it and then I can drop it from the environment and we're working and and basically treating our clusters like chatt and and that's a that's something that I know a lot of people have had a lot of trouble with um because a lot of people put a lot of effort into the development of the Clusters and all the all the Nuance that goes into it afterwards um the ability to set that up again I'm going to say it repeat repeatably and be able to say hey look the cluster this cluster is created it's created the way we need it and when I'm done with it I'm throwing it out I'm just not not going to worry about it I'm going to delete it it's going to be gone and then I'm going to use our new cluster um simply by adding and removing them from these environments so what about all that crazy nuance and all that coding and all that time and effort that I've I put into my little baby over here this cluster over here we say don't worry about it don't grow attached to the Clusters because you can drop them and recreate them at well right you can recreate them at any point right that virtual cluster capability yeah yeah go ahead um yeah we're coming up to just uh the end of the show so wanted to just kind of sum things up here and you know the snapshots are critical you know it seems like you've got an interface that allows you to you to Archive a lot of the components that you've been using so then you can pick and choose what you really need and you know what you're needing I expect there's a pretty deep search functionality behind what you offer here as well yeah AR the archiving of snapshots is is a big deal yeah yeah and it it really just shows you your whole history you know talking you know be being audited and knowing where your stuff is it's so important and it's really really hard these days so and I imagine there's a deep identity component to this so people can't just jump into any cluster that they wish to jump into yeah that's right so you know we have to support Arbok um we have we have Arbok support here and and you know we're we're putting in right now this quarter the ability to absorb uh your SSO and Sam also so you know we can follow your ad we can follow your your organization Federated um but if you don't have it our arbach is actually pretty straightforward and it's it's really nice you know especially through the CLI um everything that we do here is is you can do in the seot so we do everything just as cleanly as possible that's the whole idea so it seems to me like you know where you're going is almost like we're thinking about a transformation of what the platform actually means right so like we've become so caught up in the complexity of the platforms that it's difficult to almost take a step back and that's kind of think what you've really been trying to do here you know is take that step back and thinking about container management from a different context absolutely there there's there's a lot of conversations that a good competitive team can be having about the latest tech about what they want their their uh their Enterprise to do the the engineering discussions of how well is it running and how do we need to tweak it and where are the where are the bottlenecks and where what do we need to give attention to um and we sort of lost that a little bit when we we got buried in the in the the Myriad details um I think the ability to say there's a lot of this that is science let's remember that if it's science and we can make an algorithm for it let's get that out of our way um so we're really we're really uh pushing hard on that yeah I mean you said it like that it's really the the the the cognitive overload of how you know how how hard does it is it to get my systems and my my company to scale the way we know we should you know and so it's something like this where the status quo is cicd pipeline after cicd pipeline for each environment for each microservice for you know for each and you know we're coming in and saying well doesn't have to be like that you can actually basically archive all that information you can multiple Services right yeah yeah so the last question I have is where are you going from here and is there anything else you just would like to dis you know like to say before we uh before we end up uh yeah we're getting into gpus pretty heavy and we're getting into um you know the ability to say we can cluster gpus in kubernetes but we can also cluster gpus um in a different cluster that makes makes things wildly more efficient um that's what our customers have been asking for and so we're getting into stuff like that Mark what do you think um yeah that um we're really we've got a coup we've worked with a few um ml companies and there's a lot still to be discovered about th those workflows um and and like we were talking about earlier when you can have when you can have an elevated discussion about what's going on in that organization rather than getting mired in the details you can have you can really put yourself ahead and put yourself into better better Tech and when it comes to um the cost of a mistake or the cost of an inefficient part of the system starting to become astronomical when you're talking about gpus and whatnot a lot of those decisions become really important so what we want to be doing is speaking intelligently and looking intelligently at our Enterprise as a whole and the performance of it the scaling of it and have we chose wisely um well how are how are you going to have those conversations well you need good Tech and you need good tooling to have the right people in the room having the right discussion I mean we've all been there I just spent the last two days trying to fix this thing and it turned out to be some weird spelling of a weird thing that I had to look up in the docks like that happens and it's it's incredibly costly and and the ability to put that team on a on a footing that will um Elevate their think and Elevate their discussions um I'm really excited about it like I'm really excited to about find out what people start talking about once they're using Kodiak and what types of conversations uh they're having and and to see where that leads uh because it's it's really it's really uping the game a bit well well nice work it's pretty cool stuff thank you very much for your time Mark and Ben and we look forward to talking with you soon appreciate it guys thanks for having us it's great to be here thanks see you if you like this video please give us a thumbs up and if you'd like to see more videos like this you can always subscribe to our YouTube channel we're on all the major social media platforms you can always find us at the new stack. we hope to see you soon [Music]