How Rancher Labs is Seeing Kubernetes Put to Work in Production

Key Takeaways

I we'd like to thank the cloud data computing foundation for sponsoring our podcast from cloud nadir Khan in Seattle inspired by internet scale computing the cloud native computing foundation advances the development of cloud native technology and services by creating a new set of common container technologies and formed by technical merit and then use your value you can learn more about the cloud native computing foundation at cnc fil [Music] hey it's alex williams the new stack here launching off Kubek on cloud native days in Seattle Washington the keynote for the morning just got underway and I'm now going to have a conversation the first of the day with Elena Parker check of ranchi labs a principal software engineer Elena hello and Brian Scott managing systems engineer how are you doing Brian good how are you good so really with the big theme that we're trying to explore today is you know how how is Cooper Nettie's being used in production and some of the projects like coming out of the CNC f community and so our discussion here for our purposes really focuses on using Cooper Nettie's in production and so so Brian I want to just ask you a quick question now what are the roots of year involvement with your benetti's why did you guys start using committed Cooper Nettie's and you know and take us to maybe where we are today I think right now over the past year has been this explosion of all these different types of orchestration technologies that allow you to deploy like containers in many different ways and I think you know at Disney we've tried many different tools we were very much a technology company and you know we use a mixture of different tools in my communities kind of resonates with us I mini late on many different levels and one of them is the way that it orchestrates like containers to the actual notes so with that all the integration that it has with both private and public clouds really makes sense for us and it kind of extracts a lot of the like a lot of the issues around provisioning nodes as well as like load balancers and secrets for us so a lot of the secret sauce so we normally have to write around other tools are kind of already put like blood look like some what I provide it for us with like communities okay so how did rachel has become part of your orchestration platform in housing news okay so obviously you know we're a very large organization so we use many different types of like tools as well as orchestration methods and each team kind of has their own way of doing things so with Rancher we were able to allow our teams to use any orchestration they wanted and kind of keep everyone happy right it kind of gives everyone this like this like a single plan of glass on where they can see where everything it is laid out and we can kind of kind of get a feel for any type of orchestration that that may fit that's that death certain type of style of like application okay so late a ranger can deploy with multiple orchestration tools how's it work with cooper Nettie's as brian already mentioned you might be one orchestration tool for one of your environments another for another let's let's assume the pupae cornetta's and you want to start deploying your communities applications but before that you actually have to bring up your community system cluster and that's where it gets complicated because like you can have multiple cloud providers in the organizations and to be able to run coordinated services from multiple nodes you have to set up like the plumbing the networking the storage and we try to make it cloud agnostic and very user intuitive when it comes to the deployment so all user needs to do is to register the hospital rancher from any cloud provider AWS will lay the network overlay Network for you we pick up the storage driver and we deploy the culinary system style it's all containerized from then on your user applications deployment is delegated to Cooper medicine our work is done we continue monitoring the communities clustering if anything goes wrong we replace the coronaries containers like we keep an eye on the health checks so just to make sure that Cooper Nerys cluster is green okay so tell us about that setup you know what's involved in that set up um you know what are the complexities that you're extracting you know from the process I mean as I understand it's a master kind of note environment right and there's a lot of things that you have to do if you don't have that support right you have to be setting up the overlay Network for example yeah there's lots of take us through what your abstract incompletes atyachar abstract so we abstract it like at the hardware and them and like first you have to register your notes whether it's the clock it's a it's a virtual machine in the cloud provider or bare metal host then on top of that we start layering our infrastructure services the infrastructure services the overlay network as you said for that will let you pick NECN i plug in you want whether it's a calico we've orange your IP stack then a internal dns because like google nexus as any other service like these components have to talk to each other have to discover each other that's where dns really helps and then the health check service that's like a groundwork after all this is done you stand up your kunai disclose three communities system cluster consists of multiple services like it's ed the data plane and then the control plane API server and like yeah once this layer is set a user is all ready to deploy his applications on canary okay so tell us work Disney is right now are you in production so yes we are using our communities are using it in production um we using it with a number of different applications some of them are large plus some of them are small but I think we're some of the challenges that we've definitely been able to let tool i use rancher is to kind of cut away all of the the hard setup of committees in the past we've had to use like third-party tools as well as kind of manage our own masters and slaves within the committee's now it's kind of it it's more like it's more like of a of a like a of a like push button setup enjoy that easy yep now do you run your own infrastructure entirely yes okay and so are you what tell us about your impressions are these parents is bare metal or virtualizing environment we build utilize both on print as well as a public cloud so we're able to leverage either on print and/or amazon and or google to actually hosts are various different types of so how is yet how's Rancher extending itself to support multiple cloud environments there's now there's a sure that's available there's a AWS is not fully supportive but there's lots of people who are using AWS with ranae's what are you guys doing so furniture our our goal is to make the infrastructure deployment as well as the community's deployment cloud agnostic so the user it shouldn't be any different for you to set up like your infrastructure plumbing & lanzhou communities or messes or swarm cluster we wanted to make it as a slot agnostic as possible so it's really like one click all you have to do is line choker is cluster and you lay out all your plumbing for you okay where what are some of the what are some of the tooling that you're still looking for that you would like to see to make your operations run smoothly um so I think there's some key areas that affect almost every company within like the container space I think storage is still something that need to solve I think my secret management even select committees has a secret management I think there are certain things with that that team still have to solve for as well as dynamic configuration management of config files within your application and I think you know a role management and as well as access control it still needs the there's still need some tooling around there you know within a very large organization people come in and go and I think with that we need to be able to actually better manage who actually still has access to our infrastructure role management and access control I've heard I heard a Craig mcluckie talking about the challenges of like the developer actually still touching the physical server itself is that kind of what you're referring to or I think that's kind of with us at least is going more away through like CI and NCD right like are like we try and envision the whole life cycle starting from that first i git commit right when that gate commit is made to either github and or maybe it a get lab that kind of triggers the my whole life cycle that's the Sanitation yes iron man have you so so how do you guys treat that how do you guys treat that continuous integration from the git commit all the way through the whole kind of process itself so usually we try and make our pipelines as automated as as possible and I can kind of let let me go through kind of how that some that I'm sure integrations done so that's actually when it comes to that user application deployment and it is actually up to the up to the user in what way he wants to configure it there's lots of there's lots of there's lots of tools out there to use to do that there's there shippable there's there's worker there's different tools do you guys have your own or do you should provide a framework for like for keeping your application template so like you can you can choose whatever template you want for CI CD and you deploy it as a appliqued at the user application on the orchestration provider of your choice okay yeah yeah I think because Rancher exposed a certain API back out to the user all right we can integrate that with your drinking's or let's say bamboo any other type of ACI tool of our choosing okay so with the overall kind of like what tell us a little bit about your overall kind of you know your scale out infrastructure art are you all like you know our how are you integrating with kind of those other pieces that that you have I mean did you use red hat are you I mean I think like everyone else we use a little bit of everything it's kind of hard to say but we really trying to two little eyes as much of the api's as possible we're kind of had that tight integration so that seems to be kind of a theme that I've been hearing is like I mean the gang Craig mcluckie talked about the atom of working the API is that it I mean how do you bring that core philosophy Elena into into your interior on development of the ranch platform do you have any like how are you thinking about how are you thinking about like that automation as kind of a court ended up of a rancher provides i would say in brine chair we mostly concentrate on the automating your infrastructure and artists traders but they I think like if you look at Rancho right we it primarily focuses on infrastructure automation right and with that that allows us to be truly portable right whereas right now most solutions are either tied to either a certain cloud provider or they're tied to just unground so I think this will kind of allows us to kind of span like a multiverse of different light providers out there whether it's on print a private cloud in a public cloud we can truly be very portable with our applications so where do you want to go from here like and you know we you know with your own use of Coober Nettie's you know what's the next step what are you trying to learn here at this event particular to help you get today I'm here I mean I'm here personally to try and gather as much knowledge as possible about what's going on with communities how different companies are using it how they're tying into the into the luck amenities like API and make sure that we can actually grab as much of use out of it as possible how are ya how is it allowed yeah so what are you with the point what's the flexibility you're trying to still try to achieve where are some of the goals that you have some obviously our number one goal is delivered the best guest experience right I got to our guests right so definitely moving a lot faster is what we want to be being able to make sure that we can reflect those changes back out to our user as quickly as possible so for rancher itself the platform and you know and how it's growing where did what's the direction you guys want to take rancher we want to explore all the tools that are out there for genetics cluster deployment we want to learn from their experience we also want to follow the communities closely to see like what new features they're adding how you know how can they be integrated with Ryan share like what should be changed in the United system cluster a start-up and management so we're mostly concentrating on like on the genetic architecture just to make the user experience for communities deployment easier what new features would you like to see we are yeah we're working on the integrating with the infrastructure on kaduna is like helm catalog service that's something we wanted to which prevent your helm yeah helm then the next big feature is brunette is iterated control plane but something we want to see in venture very soon why is that just because it's it's very reliable like it's it's very good when it comes to scheduling it allows you to maintain like multiple clouds yeah kinky kinky you can configure that whole control plane in itself right yes right right and what makes that typical now it's just we need more time for that there's we just we just have to work on this integration part like we have to bring the UI we have to bring the API excuse to write your users because now is mostly just a command line now mostly like you can see you can stand up the standalone communities clusters and the grenade is just another grenadiers control plane is like one extra layer on top of these clusters so how about how about how about four-year what are some of the things that you like to see in Coober Nettie's I think Federation I just got introduced allows us to span across our applications across across on multiple clusters within my various different regions I think I still want to I think like communities still have some work to do and again in in certain areas I like storage repin merrily but for the most part it is solving a lot of people's problems so you are able to use you re your are getting a better handle on storage or staple operating solely absolute yep so what is I guess I'll ask you Bo what's your approach right now for storage i think we're trying to use you know what's currently my coming out there there's there's there's NFS there sifts for people like they are running on on on on windows people are taking advantage of the back and storage systems that are provided by both google and as well as amazon i think i mean i think ultimately folks just kind of have to be able to build their Klaus to or build their applications to better support like cloud technologies how about for you guys what's your approach to storage and danyon and dealing with stateful applications with the rancher platform in the range of clock we provide storage for brunette is using the Flex volume drivers and our goal is to add as many storage drivers as possible I made you support now we support just just a couple of them today's which EFS and Rex ray what we are thinking about any more like whatever could United supporters we should support mm-hmm the same with the networking drivers we want to add all possible C&I driver the kubernetes support interesting and I is it yeah it's not let network at all most peculiar Knight is for cuneta Africa so so do you find many customers are running Cooper Nettie's and swarm and other organization platform it seemed mostly just one or the other usually like when you when you come to write to you wanna you have an ability to try and then you make a pic but we have customers Quran coronaries and cuddle cuddles all built in our frustration puddle yeah it's called kettle but whose platform is at it it is run chair built-in toe cuddle okay actually our key street donated system cluster with cuddle yeah so but customers who use our grenadiers and swarm not really kunai this is one of the most popular artists fashion park was in Rangoon Eddie's for you guys always from the start so use everything there you doing yeah because that's what is he after day right yeah yeah it really depends on the team and kind of how they feel right and so you're using kerber Nettie's are using other platforms to that organ we use cattle you like uber at ease we're even while using missiles it's just a wide mix using swarm two isomers wall yeah right how how about over time that you know the long-term stability of these platforms are you confident in the long-term stability I think that this day and age everything's changing so quickly that you have to be able to adapt your applications to be able to pivot very quickly right so we try and make sure that we can you know at least build builder applications in a way where we can actually pivot to something new it son better let this come out about how about furniture we reinvest equally to all darkest tration platform the trying to support whether it's Carol Burnett is mezzo so swarm so it's really the user preference of what he wants to choose for his environment and even within the same company like the development environment can choose cuneta sand QA can choose masses and we support all this model have you had to create your own internal forks for instance of different platforms be able to continue to support all the different platforms overall in Rancher you mean by cuneta spork yeah or like yeah I mean like please you I mean there's a Neville looking to come to a point where like you're working upstream right in your commits maybe not are accepted not accepted or yeah and so you have to like then create your own forks internally into the platform itself is that something that you're having it is it's actually something something we started with four communities in rancho we started with maintaining our phoric because i want networking in the beginning it wasn't CNI it wasn't even see NSA and I format but we actually converted it to the CNI driver and we are working on upstream in rancho changes to kudo net is like we have a couple of engineers working on these changes but he has for now we just maintain our own genetics forth but there is nothing much rather than the our provider because urbanites release is a pretty fast and the features get in quickly and the bugs are fixed quickly so there's no real reason for us to maintain the 429 pork so for prefer for you Brian do you do you work upstream you know you know with the with the different open-source platforms or do we really wait for the the vendors to kind of you know take the lead and provide provide that service again I think because we're very strong to like a technically talented yeah the team we kind of look to open source web first yeah you know we just recently launched a bunch of open source projects within the company is out to the public and we definitely take the initiative and looking at the open source and the mixture that you can actually fit our needs and solve our problems so I guess just in conclusion I guess once again you know what're you know what do you want to see come out of no Cooper netezza Rancher over the next six to 12 months just on a high level a lot more tighter integration especially as newer features are coming out within the communities I definitely want to see some of the challenges solved around storage so what were those tire integrations be exactly then um so again what Alan was saying around la federación support okay and as well as kind of see how we can solve or at least better solve the president by storage issue around stateful applications anything specific about Federation that you'd like to see Elena what we need to do is really is just to provide the UI for Federation right it's going to be like another wrench environment good job opportunities cluster environments okay there's nothing more than just to you I management okay well thank you both for taking some time to talk with us here at Kubek on cloud native day have a great show Elena Brian thank you so much for joining us thank you for having us we'd like to thank the cloud data computing foundation for sponsoring our podcast from cloud nado con in seattle inspired by internet scale computing the cloud native computing foundation advances the development of cloud native technology and services by creating a new set of common container technologies and formed by technical merit and then user value you can learn more about the cloud native computing foundation at CNC fil [Music]