How Getting Your Project in the CNCF Just Got Easier

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 hey it's alex williams the new stack here at Kubek on / cloud native day and with us right now is dan con uh executive director of the cloud native computing foundation and chris anna check and chris leads oci which is the open compute initiative container additionally i mix I'm exit i Mick set up all the time open container initiative so Dan this is your event and it's jam-packed and you've sold out and over you know that there's more sponsors than you could rip them then you could and you could have why what's going on well I hate to reference the hype curve and and maybe curses here but we're definitely not in the trough of disillusionment quite yet this is um still riding up the side of the hype curve and folks are incredibly excited about containers and so this is our first cloud native con event right where we're pulling in cuba con Prometheus open tracing and now fluent d communities and just hopefully fostering a lot of interaction and and kind of positive cross-pollination between different groups so i have a theory about this and so perhaps we could talk about it and so my theory is that like things like docker things like Cooper Nettie's are symbolic they're also very measurable and they're very constructive type project they're very symbolic of scale out ecosystems right and building scale out ecosystems and not just building them now but managing them as well right and so as we're started and as those more companies really are i think that seems to be the driver that we see is like companies want to build scale out ecosystems right that's what they're trying to do they know they have to do it they just adjust their just grasping for any kind of information that they could possibly get around it and for years it has been done very much in an kind of an isolated cluster right there be a Twitter right right right Facebook there be linked in there be there be companies may be like on the enterprise level like via netflix or even you know Disney right we're doing these things but it is now we're starting to see almost like this like now like the pieces are connecting and scale out is what we're trying to do because really there's no choice I I think of this space as well so you know what's the meta-narrative that CMC f is trying to tell about computing and cloud and cloud computing and it's that the smart way to build modern applications is on a cloud native platform yeah and I mean there's a specific subtleties to that which is we're not necessarily saying that every legacy application you have should be ported over to Tacuba Nettie's in the next month or two and we're not so it certainly greenfield or looking at how what how you can transition brownfield and what story are we trying to avoid is that oh this stuff is all so new and so scary that I'm just going to do nothing and just sit around for a couple years because for all of our partners and the projects and everything else that would be a bad thing and so I think one of the real messages for a conference like this is just a round momentum and saying the folks making these deployment decisions are not going off on their own into a you know a spooky neighborhood that this is a it where a ton of extremely smart well informed connected companies and users are going and that you can right come along for the ride okay great so so let's get a level set here so so Chris for people out there who may not be familiar with OC I tell us what oci is and I think you know we have a central boy dance talking about what cnc f does but perhaps but you could just say more we could just say kind of like in one sentence or two you know what is OC I and what is CNC app and what is it and then perhaps you could talk to the symbiosis between the two so I know they're very separate foundations but there's also some you know there's always some relationship there so I'll kind of start with you know CNC f that the whole cloud native thing is all about micro surfaces that microservices that essentially you know are dynamically scheduled you know that run on top of containers right and so that's that's basically when we when we think cloud native and all the technology within CNC f is basically around that kind of you know mission oci is really concerned about the container aspect right defining a standard way that your containers are truly portable and will run on all the different cloud providers all the different container engines out there and so o CI is very narrowly in scope in standardizing exactly you know how container is kind of executed and also how the actually image format layout looks looks like and so that's completely it so very simple narrowly scope project that other you know taking other projects like you know criminals when CN CF cloud foundry and so on will will adopt what oci produces okay is there any different take you have on that or is it similar or what I had a journalist in Berlin asked me our relationship with CNCs relationship with OC I Chris wasn't there and so I just quickly answered we're gonna crush them under our Hugh linux foundation PR guys seeing their checking his email just like flung out of a chair like no no joking joking but I know it is very symbiotic and the the container reference architecture the container platform is just one piece of the CMC f staff it so happens that it's arguably the most important piece I am I think of in terms of there's this famous diagram the IP hourglass that talks about that helps explain some of the success of the internet and the idea is that everybody agreed on IP the Internet Protocol and then right above it you just have a couple small options of TCP UDP that somewhat is analogous to the orchestration platforms that there's you know just for open-source ones right now kuber Nettie's docker swarm nomads and mezzos and then above that it allows this incredible plethora of different options and the same thing below that you just have a relatively small number but of hardware and cloud platforms and you can live on but the magical piece is that everybody agrees on IP is your interoperability layer well you you can imagine a similar architecture for cloud native where oci is really defining that key interoperability point that allows the ecosystem to have just a myriad of options above and below and there was this funny tweet from like Kelsey recently a high tower around like you know containers should be boring right you know yeah we should all standardized on you know on you know specific approach and its really where you know the exciting you know level is really applications that's what developers really care out at the end of day so he's doing its best to standardize on the container bits right so zetta is at the hourglass and analogy with a the image format in the runtime yeah yeah I mean I mean yeah they're definitely the most critical pieces there some other interesting aspects of oci around like distribution are things that will evolve after 10 but yeah we need to kind of agree on how to you know store containers how they're laid out and how we run them and keep that as you know standard as possible so it could be reused by everyone so we don't fight about that instead innovate on the level above so now we're approaching 1 point 0 for the image format in the runtime yep how close are we we're probably I would say a month or two is probably a good estimate we're essentially rc2 for both or 10 rc2 for both the image spec and runtime spec according to how the oci governance works once we get to RC 3 which had happened a week or two the oci development you two could officially call for a vote to stamp you know stamp it as a 10 we're close but you never know things pop up you know as more people play around things made shoes but you know I'm pretty confident we'll get there in a month or two it was announced at dr. Khan in 2015 right you guys yeah yeah so ja june i think was jun 2015 the effort was officially announced in later that year we kind of got her stuff together and you know these local governance things take a while to shake out it's taking a long time to get to one point Oh is there a reason why um you know if you look at OC is a standard remember right your specifications implementation so on for standard I'd argue it's pretty fast moving if you look at other standards out there you know start working the W 30 waist and so on they take you know to take a long time so a year to come out with a 10 is not so not so shabby in my opinion so CN CF is a different kind of a is a different kind of an animal and now you are treating it almost seeming like I mean you have now you know it's a reference architecture kind of initiative you're taking on different types of projects you announced fluent d today yep you have had open tracing as part of you know joining the CNC f tell us about you know that that course at CN c f sure i mean i think the big difference is that unlike oci we're not trying to define something new we're actually going out into the world and saying what are the already successful project consuming that have already achieved subtraction with our end user base and that are that are vendors are excited about and then going to them and saying hey you should consider getting hosted at CMC f because you're not really aiming to give up anything it's still your project you're still the committee or you can still use your current processes but we're going to add in all this other value on top of it and part of that is the trying to tell cohesive story about a stack and a set of stacks that that can meet customers needs and so it's sort of not surprising that stuff can just happen faster out there because we can go find really successful projects and hopefully make them even more successful and I think fluid d is a great example of that where this is not just sort of a couple people who have this idea of oh maybe would be nice to have you know logging architecture and yeah what is fluent d tell us what it is and why did you adopt it now is that priority kind of terms the project list yeah sure so you know fluent d is simply you could view it as kind of a system to aerate logs from disparate sources so there's different you know things that emit logs and you want to kind of a great them into one place so in theory that's just you know how it works it's been a healthy you know Project X you know externally like Google you know judge Google's gcp has taken advantage of fluent e they have something like to that I think was a two thousand companies or something playing with it so it just solves a you know very important problem around lager keishon which pretty much everyone has to deal with because you're using a despair components in your infrastructure to begin with so glad to have them on board I had mentioned in my keynote Heroku as having really pioneered some some critical concepts and has and DevOps and fluid D has a great opening quote as a great quote from the Heroku founder adam williams adam something like that where he talks about this idea of logs as streams as opposed to logs as files right and that's a really critical concept in computing where you don't want to trust the file system you don't want to write to it and then find that the that container crashed before you were ever able to process it you want to admit amid these streams with the time attached to it and then be able to process them in a centralized way and then also potentially send them off to lots of different sources for different kinds of analysis and fluent d is just used by tons of different folks today and i think the statistic is that they have like 300 different connection points for different kinds of containers and stacks and projects and and such so we think it can get better I mean that we're also looking to give it some help and and we would love to feature some of the ways people are using it there are more innovative but we're thrilled to have it on board maybe it reminds me of a you know he's in there seems to be some theme among these different projects open tracing you know has obviously been developed by Ben Sigmund you know who we actually just talk to and he talked about its history you know from move dapper was yeah and how and how it evolved you know from like you know single instance environments of these distributed environments and when we talk to other people to like we talked to the people over at a pareto and how they kind of tree things and you know they're thinking of really the metadata there really that that the metadata really becomes the story almost becomes a theme that you're starting to see but it has different you know there's different kinds of their different stories to tell with that data and I don't know if that's something that you guys would agree with but it seems like there's some corollary there and the projects that we see there's a tweet that I love I forget who wrote it that in the cloud native world debugging any problem becomes as difficult as a murder mystery yeah and that's a problem if it's the case and I do think of both open tracing and flu and D as being somewhat of the solution jad to say that you need a lot of information about it's not as simple as it was the professor with the candles there's a lot more junior you want to just be able to just click in and say ah professor with the candles they need to have you need to have the DNA evidence well in but you'd ideally would like your you're tracing platform to analyze it for you and be able to tell you where the tackle is right and so that that's an aspiration but I do think a lot of what we're doing is is trying to put together some of those best practices and and show off how some of the best companies are making use of it today is my analogy to simple is it uh you know it works you know it's just funny because I never thought of it that way but yeah like I remember you know my time but you know my time at Twitter we had Zipkin you know for example be able to trace down to you know your actual all the different requests made for your service and being able to see oh you know you know the career I used at the database was was basically the laggard you know in the request so i need to go optimize this and then you go fix the problem so it's any time you have a bunch of different services tracing is critical to really understand how to how to improve things and fix things otherwise it's a bit of a nightmare so tell us about this reference architecture why you know what's the what's the what's the what's the roots behind that the room was actually pretty simple and it's a good inspiration for getting something approved which is that in the CNC f charter that was put together last December there's a quick diagram that was was written up and not no one on the TOC like the diagram and so that who didn't necessarily mean that they immediately agreed on what was what was in the diagram they didn't that they disowned the whole thing about API surfaces yeah it was very simplistic it was like buried in the schedule a not many people looked at it anyway and just it was in dire need of an update it didn't link also technologies to which is you know it didn't slot technology in the right section either so I think there's also just a not invented here and I mean these are folks who wanted to put their stamp on it and that's fine I mean I back their opinion it is kind of amazing I guess we're still vote or too short but we're very close we're close getting that approved and Alexis you're listening to this you know everyone's waiting we're waiting for that uh plus one I think the partnership with red point is is definitely missing so maybe Dan talk a little bit about yeah and so that was just a nice follow-up for this and this was a connection that Alexis put together Alexis Richardson from we've works who's the head of our our chairperson of our technical oversight committee but we consider a big job of what we're doing it at CMC f2b thought leadership and market leadership and I mean I think you need to go in just realizing that any diagram you create is going to be wrong in some ways and the people are going to criticize it in that there's no perfect story here but that it is really useful when I was a venture capitalist in two thousand i would have all these companies come in and talk about how they're just transforming this and they're going to be the central thing in this and I just I didn't even have the terminology for but the elevator pitch of I'm the uber for dog walk yeah which a person's got a company yeah I did everything but um which is yes it's super-limited but it's just taxonomy is a really useful exercise yeah because then you can start asking the more in-depth questions of oh how do you compare to the other options you know why what's your interoperability and the layers above and below and all the other pieces of it and the red point folks have been fantastic to work with on it and we're actually kind of aspirational here but but we have an idea for a real open source project behind this where it's not just a PDF or a pain but will actually be a website that and that you can go to and hover over and and then what we'd like to do is to try and remove a little bit of the judgment of what projects get included or not based on how many stars they have or if their companies how much funding they have and we haven't said all those rules up but we have some ideas for them and we're really excited that people take a look and in fact right after my keynote I I had over a dozen issues opened up all things we've done wrong we are definitely coming in which is great yeah been ignoring those yeah that so that's forming that sells kind of forms some structure about what future projects to think about and and you uh you had a whole list of them on the diagram yeah and you gotta slide too yeah we have a cockroach GP at CD i have here messaging with nats string processing with heat networking with flannel calico we've CNI configuration at CD RBC proxy with linker deal all these things that I'm and we write about all the time yep yeah so I'm so how do you prioritize you know it's it's you know the way the CNC f structured it's really up to the TOC to decide which projects get in we kind of a process in place where anyone that is interested in becoming part of CNC f will go present to the TOC so a listening people should actually get their voices absolutely it's on github they can reach out to me or file an issue and will get them scheduled I think next actually next week we have pachyderm you know scheduled without me Kyle pachyderms cool a unique little technology but you know there was a consideration they're just presenting to say okay to feel things out to see how there's a debate the presenting the TOC is really the first step okay okay it's like dating you know it's goes on the TLC so there's nine folks if you google CMC FTO see you'll get them people heard of a lot of the moon's folks like Solomon hikes from dragon dragon Hyneman for mesosphere Brian grant from Google you guys keep those those minutes are open then you can get access to everything is open okay halls all are all open every first and third Wednesday Haiti okay tific very simple okay cool that's great that's good to information to share excellent in the mailing list is also open where I encourage folks to write until they so the TOC then take it to the TSE all then listen to the wider community you know I only discussion on the mailing list and you know if the TOC feels comfortable you know with the project they'll ask them to put forth the formal proposal you know which is a process for that and then they'll be discussion around the proposal then it requires a two-thirds supermajority vote from the tsc to be accepted and you know boom you know there you go you have to accept the project Edwards yeah the other big change coming down the pike is that we're adding a new category of earlier stage projects and this was a little bit of a challenge we got into where the first project in CF CF is Cooper Nettie's which is one of the most successful highest velocity wyd Ryan's project to the last decade or so and then the second one was prometheus which is also a fantastic project I mean yeah very widely use has a huge level of adoption we think this some the format that it pulls things in can become an independent standard right and those two projects sets such a high bar that we had the fear that nothing else was going to meet it and it creates a real issue because if the only project that can come in to see ncf is sort of already wildly successful on its own then CN CF is not adding a ton of value but on the other hand you know the technical Oversight Committee those people feel strongly about their reputations and feel that their vote represents something and so they don't want to necessarily support piddly little projects or sort of uncertain ones so we're trying to thread that needle by creating this new category projects that we're likely going to be calling inception stage and that's hopefully it's going to be a new process that gets approved in the next month or so and what it does is allow more less future projects to join cnc F at an earlier stage but unlike incubation projects where when they're in they're basically in maybe it'll take some number of years for them to complete the work to graduate but it's basically a sure thing with the inception stage is set up so that every 12 months it requires a new vote of the TOC and that vote can either maintain them in Inception for another 12 months if things look promising it can move them up to incubate it or all even all the way to graduated but if there's not a supermajority to keep them they actually will be what will lead the TOC believe is how does the TSE chosen it's kind of complex yeah I mean initially it was based on there was five from the original governing board were we're notice was at six or five I'm trying to recall now it was it was six the goat initial governing board chose six financials cnc app governing board yeah that was the initial governing board win-win bootstrapped the and then once those are the original members yes yes and the terms are pretty long so the terms either two or three years depending on you know how things were done and then we allow the TOC itself the freedom to elect to more people so from these initial six that were bootstraps they got you know they chose two more members and then from kind of our burgeoning and incubating and user community we were able to choose a representative to be on the TC to TOC to represent the end user community interests so that's how we got to our 9 and their long-term so you know we're talking three year to your terms what happens when the terms end spots available for a new wave of I think grandfathered in or did they or is their term limits or no there's term limits it's all it's all in the toc github haha you know read me I'm just curious on how these things work yeah yeah it's uniquely structured I would say though that it is somewhat wouldn't quite say unique because really the model for it is the Linux Foundation okay where you've had the vendors on the board and you have the technical advisory board yep of kernel developers and their elected every year at the colonel summit by the top colonel contributors no and then what the chairperson of the tab serves on the Linux Foundation board so I'd say that the Linux Foundation has really focused on this model of trying to separate technology decisions from business decisions and we feel like this has really served us well over the last decade and cnc f was emulating that we had a little bit more of a challenge in that we were both who'd strapping it and it so just a wider range of technology so we've covered a good bit of ground here you know anything anything in finale that you guys would like to share I'll just do two quick pieces that came up in my keynote one were excited to be launching this kuber Nettie's managed service provider probe and a training and certification program that goes with this and that will include a free massively online massively open online course MOOC about Cooper Nettie's and then that can folks can sign up for a training it's going to be one hundred ninety-nine ninety-nine dollars right now for people who sign up early that can then lead you to become a certified Cooper Nettie's professional so for folks looking at this space I think that's a neat opportunity and then the other one is that we have drastically dropped the end-user pricing so that it is now for start-up the same cost as buying two tickets to cloud native con q con for a big company it's the same as buying five tickets and you get those tickets as well and so our goal is to get a lot of companies that are using these technologies in different ways in a real diversity of companies geographically size the company what kind of business there in what clouds they're running on what kind of metal all those things and to just have a lot more diversity around feedback in and so if that's at all appealing to your your listeners I urge them to sniffing just I know you know I have two things also one you know like I mentioned before oci is getting really close to that one out so if you have a stake in the you know container space please get involved and give feedback before we kind of cement things would be greatly appreciated and also you know from today's keynote we announced the events for next year for you know cnc f's Oakland native con Europe is happening in march in berlin and also the one in North America will be in early December in Austin so if you're interested in sponsoring please do so early since we sold out of sponsorships you know this year and i'm assuming next year it's going to you know happen and same with registration so register early submit your talk cfp is open at least for the one in berlin and just do it otherwise you know we may be another cell that situation so hopefully not the drinks ourselves but we'll see how things go well great well thank you guys for participating in dan thanks for joining us for we're pancakes this morning yeah I hope we're gonna be to do the same thing in Berlin in Austin that'd be fun great thank you well thank you guys 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 you