#134 January 19, 2021

CNCF and the Linux Foundation, with Chris Aniszcyzk

Hosts: Craig Box, Adam Glick

After building the Eclipse IDE and Twitter’s Open Source office, Chris Aniszcyzk bootstrapped the CNCF, joining its parent the Linux Foundation in 2015. He’s now a VP of DevRel there, as well as CTO at the CNCF and Executive Director of the Open Container Initiative. Chris joins us to share his technology journey and Cloud Native predictions for 2021.

And all that is now
And all that is gone
And all that’s to come
And everything under the sun is in tune
But the sun is eclipsed by the moon

Do you have something cool to share? Some questions? Let us know:

Chatter of the week

News of the week

  • ZX Spectrum

  • GORILLA.BAS

  • Gentoo Linux

  • Java Virtual Machine (JVM)

  • Eclipse

  • Backstage and Roadie

  • Twitter OSS

  • Linux Foundation, and its sub-projects CNCF and OCI

  • Chris’s Cloud Native predictions for 2021

  • Chris Aniszcyzk on Twitter and on the web

    Transcript

    Show full transcript

    CRAIG BOX: Hi, and welcome to the Kubernetes Podcast from Google. I'm Craig Box.

    ADAM GLICK: And I'm Adam Glick.

    [MUSIC PLAYING]

    CRAIG BOX: Some sad news to share.

    ADAM GLICK: Yes, this will be my last week with the podcast. I'm excited to be moving on to my next adventure. But I will be leaving Google. Craig will continue on the podcast without me, hopefully with a new host. One of the wonderful things about this community and the open source world is that things that are built live on beyond their creators. Other people get involved, and they continue to do great things. It has been my true honor and pleasure to have done this for-- gosh, what 2 and 3/4 years at this point--

    CRAIG BOX: At least!

    ADAM GLICK: --bring this to all of you, to be involved with the community, and this was our way to help give back to a community that we love so much. So thank you for allowing me to be part of your lives as you've listened to this. And, Craig, this has been absolutely one of the highlights of my entire working career is getting to do this with you. So thanks to you as well for being a great host.

    CRAIG BOX: Thank you, Adam. And the same to you. Doing the show with you every week has been an absolute pleasure, a career highlight for me also. Just for people who don't know, Adam was really the driving force in getting the show off the ground within Google. I couldn't have asked for a better co-host. I'm very proud to say that he's become a wonderful friend. Sometimes people write in to thank what they assume is a giant team who put the show together every week.

    ADAM GLICK: [LAUGHS]

    CRAIG BOX: We have a little bit of help with booking and editing. But pretty much everything else related to the show has Adam's my fingerprints all over it. And his departure, obviously, is going to be a great loss for the show. I know my mom will be heartbroken. I think she prefers your voice to mine.

    ADAM GLICK: [LAUGHS] I'm sure she loves you more. It's OK.

    CRAIG BOX: But again, congratulations on the new gig. I know it's a fantastic opportunity. I look forward to seeing continued great things for you in the future. For everyone else out there, please keep in touch with Adam. You can find him on LinkedIn. Please continue to listen to the show. We'll be bringing you a few episodes with some guest hosts, some names you may have heard in the past, and we'll have an update on how we take things forward. But ultimately, it should be very much like you used to, just without the funny-sounding American guy.

    ADAM GLICK: [LAUGHS] Thank you very much. And thanks to all of you for listening. Shall we get to the news?

    CRAIG BOX: Let's get to the news.

    [MUSIC PLAYING]

    CRAIG BOX: Dutch start-up, RedKubes-- with a K-- has launched a community edition of its Otomi Container Platform and taken 1.2 million euros in seed funding. Having pivoted from a managed Kubernetes service to an add-on for any hosted version, Otomi provides one-click installation for add-ons on top of Kubernetes, such as Prometheus, OPA, Harbor, Istio, and Knative, from a web console. It then integrates monitoring and logging, CI/CD, and policy management. RedKubes suggests you may prefer independence above and beyond what your provider offers and hopes you will eventually move up to their Enterprise Edition.

    ADAM GLICK: With Google's Anthos general availability on bare metal, Hyper Converged Infrastructure, or HCI vendors are starting to integrate it into their product portfolios. This week, Nutanix announced their HCI now runs Anthos without requiring a separate hypervisor license.

    CRAIG BOX: Last September, VMware announced the separation of their Tanzu product into different editions. And now an advanced edition is generally available. All the value-add technologies for developers and operators are now part of this edition, including build services, service mesh, and integration with the spring framework. VMware also announced the rebranding of Pivotal Labs to Tanzu Labs.

    Along with the closure of Pivotal Web Services this week, VMware has purged almost all remaining remnants of the Pivotal brand. It took less than 18 months for the Tanzu brand to subsume Pivotal, which was almost 32 years old. The name still lives on in the Pivotal Tracker project management software.

    In hiring news, VMware is looking for a new CEO after Pat Gelsinger announced he would return to Intel on February 15, the company where he previously worked for over 30 years. Send your applications to us, and we'll forward them on.

    ADAM GLICK: In GKE this week, the container storage interface driver for Google Cloud persistent disks is now generally available. Previously, you would connect your disks using the entry GCE specific code. This new code path allows you to take advantage of the new PD features without having to manually manage the CSI driver lifecycle. Google Cloud operations for GKE-- formerly Stackdriver-- now also includes new modes, where it offers only monitoring or only logging as opposed to requiring both.

    CRAIG BOX: Founded by the creators of the DockerSlim open source project, a new startup called Slim.ai was announced this week. They launched with $6.6 million in seed funding to build a container devops platform. In this case, the AI stands for Application Intelligence, using static and runtime analysis to build and optimize containers. The platform integrates with CI/CD pipelines, container registries, and source repositories.

    ADAM GLICK: Grafana Labs hosted product, Grafana Cloud, was updated this week with a new free tier and more generous limits for the paid plan. Three team members can now access 10,000 time series and 50 gigabytes of logs for up to 14 days at no cost. The pro plan now has five times the previous storage for $49 a month. And you can, of course, call for custom pricing if your requirements exceed those limits. Learn more about Grafana Labs in episode 122, with Torkel Odegaard.

    CRAIG BOX: Sysdig has released their fourth annual container security and usage report, skipping 2020-- don't we all wish we could do that-- and labeling this report 2021. Insights included show that 74% of organizations are scanning container images in the build process. It's not all good news, though, as 58% of those containers run as root. Have we learned nothing?

    Interestingly, 21% of all containers run, according to survey respondents, live less than 10 seconds. Advertising included shows a 300% increase in the adoption of Sysdig's open source project, Falco, which can mitigate the impact of bad configuration with runtime scanning. You can learn more about Falco in episode 91, with Leonardo Di Donato.

    ADAM GLICK: Listener Alvaro Hernandez wrote in with a 'just because you can, doesn't mean you should" blog post. He rented a bare metal machine from AWS and used it to run a 63 node Kubernetes cluster using Firecracker. Having tried 100 MicroK8s nodes, 63 it was the most he could get running. That sounds like a challenge someone else might want to accept. If you can get more nodes running on a single box, let us know.

    CRAIG BOX: Finally, a self-proclaimed definitive guide to Vertical Pod Autoscaling from long-time listener, Povilas Versockas. VPA provides an automatic way to set a container's resource requests and limits using historic CPU and memory usage data. It's available on open source Kubernetes or in an enhanced edition on GKE.

    ADAM GLICK: And that's the news.

    [MUSIC PLAYING]

    ADAM GLICK: Chris Aniszczyk is the chief technology officer of the Cloud Native Computing Foundation and Vice President of Developer Relations at the Linux Foundation. Welcome to the show, Chris.

    CHRIS ANISZCZYK: Glad to be here. Long time listener, so happy to finally have an opportunity to speak to all y'all.

    CRAIG BOX: We probably all had a similar path into technology. What was your path to Linux and open source?

    CHRIS ANISZCZYK: It truly started when I was a younger child. Just like anyone in our youth, we played video games. One of my first pieces of computing was ZX Spectrum-- or "zed-ex" Spectrum, however you want to refer to it-- where I learned how to play R-Type. There was this crazy jetpack game that was cool that you fly around with a jetpack. Got a 486 machine that my dad got me-- awesome-- 66 megahertz, I believe, at the time. And part of that experience was there was this ridiculous gorilla game that he showed me, GORILLA.BAS.

    CRAIG BOX: QBasic!

    CHRIS ANISZCZYK: Yeah, QBasic was the intro to programming. And with GORILLA.BAS, you could open up GORILLA.BAS, and here's all the source code. I could go mess with how the physics works, and so on. So that was really kind of the first foray of, oh my gosh, I could make changes to this source file and change how this video game worked. And that got me hooked into programming in general.

    And eventually, as I got older, high school times, we had a Vax cluster--

    CRAIG BOX: At your high school?

    CHRIS ANISZCZYK: At the high school, yeah. Connecticut was a very nice privileged upbringing. And so I learned a little bit about DCL-- I think it was the programming [CHUCKLES] language at the time-- so it had a lot of fun things. And eventually that led to experimenting with Linux. At the time it was Slackware. I eventually discovered Gentoo. I had this thing where I was just like, this is amazing, compiling everything from scratch.

    [LAUGHTER]

    CRAIG BOX: That's an interesting definition for amazing.

    CHRIS ANISZCZYK: The cool thing about Linux in general-- really why I'm involved in open source till this day-- is when I was first getting involved, I couldn't get sound to work. I was trying to get the ALSA driver--

    CRAIG BOX: Did you try Windows?

    CHRIS ANISZCZYK: [LAUGHS] I literally jumped onto IRC and could just ask people, hey, I'm having issues configuring this ALSA driver. And people from all over the world just pitched in and answered questions. There's nothing there. It was just like, here's an answer, hope you could solve your problem, and I just got hooked to that amazing community feel of people just answering my random esoteric Linux questions.

    And eventually that led to me becoming not only a user, but I was a maintainer. I maintained a lot of the JVM-related packages on Gentoo for probably roughly five years before I got a full-time job. I didn't have time to compile my kernel [CHUCKLES] packages from scratch every day.

    CRAIG BOX: So for someone who's obviously come from gaming and gone through high school, why Java? Why the JVM as your packages of choice?

    CHRIS ANISZCZYK: That really became through some internships. One of my first professional internships was at an insurance company doing Java stuff. And so that's how I learned it. Eventually, one of my co-ops of university was with IBM. IBM, obviously at the time, was a big Java shop.

    At the time, there were a couple of things that they were doing. One was trying to fit the JVM on-- let's call it the home of the future-- in small devices, like, hey, can we get JVMs running everywhere and sensors? People were calling it smart homes back in the day.

    CRAIG BOX: The refrigerator of tomorrow.

    ADAM GLICK: Like X10 kind of things, over carrier current, that kind of stuff?

    CHRIS ANISZCZYK: I think they were just too early. They had this lab of the future, they called it. And so I got involved with JVM-related things at that time. It eventually led to a long storied history with something called Eclipse.

    It was kind of the IDE that IBM acquired out of a company in Ottawa called OTI, essentially VisualAge folks-- if you remember VisualAge back in the day-- and they built Eclipse basically to simplify how IBM did internal development tooling and ensuring that they were not replicating building IDEs for each specific product, and so on. And so I got heavily involved with that community because I just love building developer tools. It's a lot of fun working on something that you get that immediate customer feedback.

    CRAIG BOX: You say immediate, but your IDE's written in Java.

    CHRIS ANISZCZYK: [LAUGHS]

    CRAIG BOX: Do you mean eventual customer feedback, over time maybe?

    ADAM GLICK: You put it into an intermediary language. And then eventually you get the feedback compiled for you [CHUCKLES].

    CHRIS ANISZCZYK: In particular, Eclipse is pluggable just like Mozilla is in Chrome, with extensions. And so I was mostly responsible for what we called the plugin development environment, for other developers building extensions. And that feedback loop was super quick because developers would just get super pissed if you made small changes.

    I remember one time, we went from, essentially, [CHUCKLES] square corners to beveled or curved corners for each file tab. And people just flipped the hell out, sent nasty-- at the time, it was Usenet messages-- newsgroups-- [CHUCKLES] to the developers.

    ADAM GLICK: You referred to Eclipse as a simplified developer experience.

    CRAIG BOX: I'd hate to see what a complicated developer experience looks like.

    CHRIS ANISZCZYK: [LAUGHS]

    ADAM GLICK: The Eclipse experience today feels a little bit like the open source version of Visual Studio, which is a very feature rich, but also compare that to some of the things like Atom or VS Code these days, or people that use "vi", what was the experience like that was the simplified version?

    CHRIS ANISZCZYK: If you recall, back in the day, developer tools-- you have your "vi", EMACS folks--

    CRAIG BOX: QBasic.

    CHRIS ANISZCZYK: Yeah, [CHUCKLES] QBasic. If you're going to start debugging applications, like I'm going to start a screen session or whatever-- I'm going to fire up GDB over there and-- debugging was not a solidly-integrated experience. The whole idea of Eclipse and integrated developed environments in general was mostly all around, hey, can we have everything that you require to debug, develop, maintain applications, in one simple experience, supporting multiple languages too-- so not only just Java, but C, C++, Python, Perl, all these things-- into one simple development experience.

    These days, I'm mostly a VS Code person. Eclipse is near and dear to my heart. But VS Code is a solid piece of engineering. And I do feel better that a lot of the original eclipse people-- folks like Erich Gamma, one of the Gang of Four folks that helped steward a lot of the OG Eclipse stuff at IBM went to Microsoft probably a decade ago, charged with the mission to build a web-based IDE. So I feel good about that.

    CRAIG BOX: Now I don't want to let slide something you said before, Adam. You equated--

    CHRIS ANISZCZYK: [LAUGHS]

    CRAIG BOX: --VS Code and "vi" as being roughly equivalent in terms of the complexity.

    CHRIS ANISZCZYK: [LAUGHS]

    CRAIG BOX: VS Code needs an entire JavaScript environment just to write "hello world"! Come on!

    ADAM GLICK: It's a spectrum.

    CRAIG BOX: But not a ZX Spectrum.

    ADAM GLICK: That's been a pretty big shift that I've seen. So I was curious when you defined it as simple, what the experience was, because you mentioned developers give a lot of feedback.

    CHRIS ANISZCZYK: Yup.

    CRAIG BOX: They're the kind of people who know that they could write a better IDE if only they had a spare weekend.

    ADAM GLICK: How do you deal with the fact that you're getting a lot of feedback, but there's also a lot of people who can contribute, and certainly a lot of strong opinions about everything between tabs and spaces and coloring, and then also what are [? supported ?] the languages?

    CRAIG BOX: Roundness of corners.

    CHRIS ANISZCZYK: Yeah, [CHUCKLES] roundness of corners, and so on. My lesson during those times was trying to be as transparent as possible with the roadmap and allowing people to get involved in what they truly cared about. I do think you still need a very small core set of folks to make the decisions at the end of the day. I love that experience. But I don't know if I would ever be in a position again where I would want to go build a developer environment that so many people depend on.

    I sympathize with the VS Code folks these days, because it's becoming the de facto IDE. If you look at the Stack Overflow survey results, and if you see common IDEs for all the languages, it's almost like VS Code is number one or number two for all them. And that's just an incredible amount of communities that you have to support.

    I also think pushing things down to a plugin level or extension level-- so if someone writes a Rust-based thing for VS Code, that community can own it, not necessarily people who own the extension infrastructure.

    ADAM GLICK: I'm seeing some parallels between that and how Kubernetes has chosen to expand out and other projects. How do you scale without creating giant monoliths?

    CHRIS ANISZCZYK: Correct. Yeah, that's the fundamental lesson, trying to decentralize and distribute decisions as much as possible through an extension mechanism, whether it's plugins, Chrome extensions, or CRDs, for example. I think it's a key lesson for scaling open source projects.

    CRAIG BOX: There's another aspect to this which I find quite relevant today, which is that you are working at a company who as a side project, happened to have a piece of technology. And then you went and worked with some people from that company who decided that they were going to leave and found a company to work on that full time. And then you follow through the Red Hat and then also will touch maybe a bit on the Eclipse Foundation. How's that journey relate to what you see people do today?

    CHRIS ANISZCZYK: Today, I think it's very common for people to develop a piece of software at a company, whether it's Twitter or Facebook or some other large type organization-- then that piece of open source software gets popular, and then you're like, huh, maybe I've worked here for four or five years. Maybe I want to do a startup around it. Or VCs approach people to do a startup. So I think it's a very common project that's produced at a larger company becomes popular, employees decide to potentially go do a startup-- I think it's a cycle that we see all the time.

    We've seen it in the Kubernetes ecosystem plenty of times with folks like Craig and Joe, who did Heptio. We're seeing it a lot in the CNCF ecosystem. I even came across a startup in recent weeks around Backstage-- which is a developer portal. And there's a startup already around it called Roadie. It's just a common-- I think it's a good thing. Not everyone is going to work at a company forever. Millennials these days jump companies, what-- average tenure in Silicon Valley these days is like 1.5 to 2 years. I think you just have to build in for that.

    CRAIG BOX: I feel like a dinosaur.

    CHRIS ANISZCZYK: [LAUGHS] In my day when we were at IBM early-- before open source focused startups was popular, we were there about five years, a lot of us worked on Eclipse-- we were approached by smaller companies that wanted support, either, hey, build us a custom plugin or, hey, can you build us a product around something? IBM at the time, wouldn't care about these customers because maybe the revenue they would get from them is under a billion dollars.

    ADAM GLICK: Too small?

    CHRIS ANISZCZYK: Yeah, too small, so they would just ignore them. Eventually, we were like, enough people are pinging us-- and at the time I was also finishing up business school at the University of Texas at Austin, where IBM graciously at the time, had a program where they would pay for business school. I don't see that often these days. But that was an awesome thing that I'm super thankful for. And my business school brainwashed me or convinced me to, let's do a startup. This is great. It's going to be easy.

    [LAUGHTER]

    So we did that. Eventually, a couple of us left. And we still maintained all these Eclipse projects. And then a week after we left, other IBMers at the time we're like, what's going on? What are you doing? You're basically doing the same thing you're doing, answering questions and so on, but actually not getting paid for it now-- was the joke. I'm like, yeah, it took us about three months [CHUCKLES] to actually get our first revenue.

    But it was a fun experience. You learn a lot, especially if you put yourself in a challenging situation like doing a company from scratch. So I was super thankful for that experience.

    ADAM GLICK: So where did you head after the startup? You didn't quite become the unicorn.

    CHRIS ANISZCZYK: No. We did OK. So what was interesting at the time, is we had a lot of companies that were using Eclipse at scale, folks like Cisco-- even Google at the time was one of our customers, where I remember to this day visiting Mountain View, and you're like, oh, Eclipse is a little bit slow.

    CRAIG BOX: Maybe it shouldn't be written in Java.

    CHRIS ANISZCZYK: Yeah. It was like thousands of JARs in the class path. And they're like, oh, we have another approach. What happens if we do a mega JAR or a fat JAR? We'll throw them all in to make it faster. So it was a lot of fun.

    So eventually what happened was Google ended up buying one of our competitors because they were building-- I don't know if you remember this, I guess we're all a little bit old-- Google Web Toolkit, or GWT-- remember this at all?

    ADAM GLICK: Mm-hmm.

    CHRIS ANISZCZYK: You would write Java, cross compiled a JavaScript, so you could use Eclipse to actually debug your JavaScript because at the time all we had was Firebug. We weren't blessed as we are today with Chrome DevTools and being able to do breakpoints from your browser. So they eventually ended up buying that company there. And we decided just to go sell off the company. Not a fire sale, but we sold it off.

    And I eventually went to Red Hat, because I wanted to work on open source full time. And Red Hat was like, come, please, make JVM and Eclipse related stuff better on Red Hat Linux, and we'll pay you to do that.

    ADAM GLICK: You went IBM to Red Hat before Red Hat went to IBM?

    CHRIS ANISZCZYK: Yes.

    ADAM GLICK: [CHUCKLES].

    CHRIS ANISZCZYK: I had the unfortunate or fortunate experience of being at both of those companies [CHUCKLES] experiencing the cultures. And I will say, it's been interesting to watch how they have merged and allowed each other to have enough freedom to go about doing their business. So definitely different cultures, but I think they've been fairly successful giving Red Hat at least enough space to be successful. I truly believe Red Hat is a great fantastic place to work as an engineer. They truly do care about upstream communities. But yeah, a fun experience-- don't recommend it [CHUCKLES].

    ADAM GLICK: So, you spent a lot of time in open source and companies building open source tools. And then you shifted to the internet service world. What drove that shift?

    CHRIS ANISZCZYK: When I was at Red Hat, at the time-- I loved my experience, we had a great team-- I had a couple of colleagues that were reaching out to me. Google was doing some interesting things. A couple other Silicon Valley companies-- eventually Twitter reached out. In 2010, we started to have conversations. And they were like, so I'm familiar with the service. But why do you need my skill set in particular?

    A couple of things. One, we're in the process of basically moving off of our Rails, Ruby codebase to the JVM. Obviously, I had a lot of JVM experience. And two, which was a little bit more interesting for me, was, we're looking to basically figure out our open source strategy in general. We've just randomly open sourced things, had some issues of not complying with licenses properly-- we just need someone to own this problem. Ideally, we'd love to have something similar to what Google does with Chris DiBona's office.

    And so I'm like, that's kind of cool, help some engineers with JVM stuff and also do some open source related things. This sounds like a fun opportunity. And we argued for about 6 to 12 months about the relocation to the Bay Area. I eventually sucked it up and moved over, enjoyed a couple of years in San Francisco.

    It was a crazy time. It was a typical startup growing significantly. It had a lot of interesting infrastructure problems, which I learned a ton from. These days, we're blessed CNCF has all these projects, Apache. At the time, there was really no open source orchestration options.

    ADAM GLICK: You had to roll your own for a lot of those things.

    CHRIS ANISZCZYK: You had to roll your own. And we had-- at the time, though-- it was funny-- the company was about 25% ex Googlers too on the engineering side. So a lot of people were replicating stuff they had at Google. So it's like, we want Blaze. We don't have Blaze. So we're going to build something called Pants [CHUCKLES].

    CRAIG BOX: [LAUGHING] I can see how you get from Blaze to Bazel. How do you get from Blaze to Pants?

    CHRIS ANISZCZYK: [LAUGHS] I don't know where the original name came, but it's ridiculous. The commands are like, pants clean, pants wash, and you're like, whatever, folks. But Mesos came out of that.

    The company basically really truly wanted to open source a lot of the infrastructure it was building, because at the time it decided, look, we're in the business of advertising, software is just a means to an end for that, so let's just open source all we can that's not related to our ads business, so we could attract great engineers and potentially get ahead of the market because a lot of stuff at the time you couldn't just buy off the shelf. You had to go create your own Redis, or memcached clustering mechanism and so on.

    So we shared all that. It was a great experience. And what's interesting to this day-- I'm very torn because Mesos holds a special place in my heart, and I helped create MesosCon, I helped built a lot of that community-- when the Kubernetes was getting popular, it's interesting to contrast how those communities competed and built out over time-- probably worthwhile for another long podcast. But it was a great experience.

    ADAM GLICK: But how does the community view you given your current role of--

    CHRIS ANISZCZYK: [LAUGHS]

    ADAM GLICK: --are you somewhere between Benedict Arnold and Judas? Or--

    CHRIS ANISZCZYK: No.

    [LAUGHTER]

    We're all friends. Ben Hindman, a lot of the Mesosphere-- I guess D2iQ is the new name. I think that the lesson there is if you're truly trying to build an ecosystem that involves multiple winners, and so on, I think it's important sometimes to give up a little of that control. And the fundamental issue I think with Mesos, maybe it was a little bit too tightly controlled by one organization. They were not able to diversify properly, where I think Kubernetes did a fantastic job.

    A lot of the initial leadership-- a lot of people were just very mindful about that and just built a better, more open, and inviting ecosystem. Even though at the time, Mesos I think was just a better technology and so on-- you could argue about two-level scheduling versus one-level scheduling and so on-- but wider ecosystems win at the end of the day.

    CRAIG BOX: So on the topic of wider ecosystems-- you are now, of course, at the Cloud Native Computing Foundation-- how did you get from there to here?

    CHRIS ANISZCZYK: So I was at Twitter for roughly five years and got lucky. You join a company early, it goes through the typical Silicon Valley dream of going public and so on, so I was benefiting from that experience. But after five years of roughly going from little under 100 engineers in the organization, and then five years later it was about 2,000 engineers, I was a little bit burned out, to be honest. And I was like, I got a great opportunity here, a great exit-- super fortunate. I'm going to take some time off. Maybe travel the world for three months or something.

    And during that experience, I got an email from Jim Zemlin, who runs the Linux Foundation. And he's like, hey, Chris, we're looking to start a foundation around Kubernetes. You have a lot of experience, obviously, with Mesos and all this container stuff. And we're also starting an organization around doing container standards-- Open Container Initiative, which a lot of people don't know about, but played a very important part in the early days of CNCF. He's like, are you interested in helping build out and run these two organizations, kind of scaling them out?

    And I was like, the non-profit route sounds interesting. I don't think I'm ready to jump into a big company at the time or another startup. So why not? Let's try out this non-profit opportunity. Sounds great.

    I've always wanted to be involved in a foundation. I've been involved in the Eclipse Foundation for nearly 15 years. So I'm like, this is great. This is an opportunity to start something on my own. The downside is it turned out to be just as bad as a startup because it just grew crazy.

    [LAUGHTER]

    The first year at CNCF was very relaxing. It was actually not that bad. It was very simple. We were just getting things bootstrap-- for OCI, if you remember, we jokingly called it the container wars. It was a mess. The first year of OCI was just brutal.

    And then CNCF picked up the year after that. I was just looking for a break. A lot of us work at these companies [CHUCKLES] where sometimes you get burned out. You need a break. Redoing engineering ladders three times through Twitter is sometimes too much. You need something a little bit smaller.

    CRAIG BOX: Condensing it down to 140 characters at a time.

    CHRIS ANISZCZYK: Yeah, or now 280.

    CRAIG BOX: Level three, writes code. Level four, writes code good.

    CHRIS ANISZCZYK: [LAUGHS] Yeah, exactly. I'm sure you go through the same experience. How do you rank each other? Do we do a scale of one to four, pass or fail? It's crazy. But every company goes through this. You get burned out after a while. And everyone needs change at the end of the day. I think it's healthy to do new experiences and join new companies.

    ADAM GLICK: These days, you're the CTO of the CNCF.

    CHRIS ANISZCZYK: Yup.

    ADAM GLICK: What is the job description for that?

    CHRIS ANISZCZYK: I started the organization as basically employee number one, started out as bootstrap executive director. The organization, we got it off the ground. It started to grow like crazy. We eventually ended up hiring the late Dan Kohn to run the business side as executive director. But I prefer working with technical communities, cultivating them, growing them, and so on. And so as CTO of the organization, I'm essentially responsible for ensuring that our projects and maintainers are healthy and happy.

    My team essentially works on advocacy. We help projects put on events. We do technical writing for projects. We've built great sites for etcd, kubernetes.io -- we helped it move to [the Hugo theme] Docsy. And so we do a lot of services for projects, ensuring that they're fully happy, which is a little bit unique for an organization. A lot of open source foundations generally are fairly lightweight in terms of staffing and support.

    Something with CNCF that I want to do is actually provide project support because a lot of times projects are comfortable just working on features and bugs. What do maintainers hate working on? Docs, advocacy, marketing-- which a lot of developers don't see as important, but I actually view it crucial. To me, it's just as important as the code and a great way to involve people that are non code or non-technical roles into your project.

    ADAM GLICK: That is completely true. And I think you're highlighting one of the important things that the CNCF does for the community that not everyone thinks about, is all the things that you need besides just creating the code to have a successful project. But a successful community has all of these people working together on these different disciplines building something to make it great.

    CHRIS ANISZCZYK: Yeah, 100%. And kudos to a lot of the original Kubernetes folks that were also very much on the same page. Sarah Novotny did a great job in the early days of pushing for this. And so it's very easy to help convince other parts of the CNCF leadership and board that we should invest in this area, because there's lots of fun conversations in the early days of, hey, whether we should pivot to be completely just be Kubernetes foundation, or should we be more of a tent and more inclusive of other projects? So fun discussions during those days.

    CRAIG BOX: So what things do the CNCF not do by choice?

    CHRIS ANISZCZYK: We generally do not hire people full time to do feature work. For example, generally anything that would be blocking a release or critical feature work, we don't do because essentially that ends up competing with our member ecosystem. CNCF, we're a humble non-profit.

    We do well, but we cannot compete with the trillion-dollar clouds out there for talent. It's just not at the same scale. Unless if you're a hippie that wants to work at a non-profit, perfect. We got that market. But it is very difficult to compete with GSUs and all these folks. So that's generally what we do.

    And honestly we don't think that's fairly healthy. And the way that works is if you look at the Linux kernel itself, Linus and Greg KH work for the Linux Foundation. They're Fellows. But if you know the Linux governance process, it's BDFL model, Benevolent Dictator For Life. Linus controls things.

    ADAM GLICK: [LAUGHS]

    CHRIS ANISZCZYK: And so having that person be employed by independent foundation versus a company helps enable trust. Luckily, Kubernetes is distributed, open governance, there's multiple leaders, rotating steering committee, and so on, so you don't need that. And us hiring people to work full time on Kubernetes for feature work doesn't make sense.

    But documentation, advocacy, that's completely OK. There's just less competition on the member ecosystem side for those types of hires. And people are generally very thankful if anyone contributes docs to a project.

    ADAM GLICK: One of the people that we've wanted to get on the show for quite a while and haven't had a chance to yet is Linus Torvalds.

    CHRIS ANISZCZYK: [LAUGHS]

    ADAM GLICK: You are at the Foundation. Have you had a chance to work with him? What's he like?

    CHRIS ANISZCZYK: Oh, Linus is great. He's super humble and quiet in person. At KubeCon China, I think last year, we actually went on this crazy panda tour, where we saw pandas and dressed up in these little crazy costumes to see the pandas. And Linus is a super chill person. Online, he's probably a little bit more forward and direct--

    ADAM GLICK: [LAUGHTER]

    --given his Finnish roots, but always happy to answer questions and advice. I think both Linus and Greg KH-- who's number two in the hierarchy-- both are super receptive to questions. And in the early days, when Kubernetes was starting, we actually brought in-- I think Greg KH and tried to get Linus at the time-- to give advice on scaling governance. If you're interested in having Linus on the podcast, I could see what I could do.

    CRAIG BOX: Put in a pull request.

    CHRIS ANISZCZYK: Yeah, [CHUCKLES] I think Linus is perfect for Twitter. But there's just been [CHUCKLES] some issues in trying to get him here. He's not a fan. He was a big Google+ guy back in the day-- huge into the Google+.

    ADAM GLICK: Oh, wow.

    CRAIG BOX: We're sorry for breaking his heart.

    CHRIS ANISZCZYK: I think he was actually using it. But I don't think he's on Twitter because the brevity is too much. And he's a little bit more long winded.

    CRAIG BOX: You're also the VP of DevRel at the greater Linux Foundation. Was that something that came up after the CNCF had run for a while? Or was that something you'd been brought on to do at the beginning?

    CHRIS ANISZCZYK: It was in the beginning. So one thing that many people don't realize is the Linux Foundation is essentially a network of foundations. There's things like CNCF, OCI, LetsEncrypt.org-- free SSL for everyone. They're fantastic. I love those folks-- RISC-V foundation-- so I was brought in to help figure out how we manage those communities. How do we keep developers across those communities happy? Which tools and services we should provide, and so on.

    So if you look at a lot of the stuff that we've built, we have this new platform called LFX, which we've launched. And it's essentially just tools that we think are useful for projects. So for example, running internships, we're huge fans of Google Summer of Code at CNCF and the Linux Foundation, but we've had a lot of requests for projects like, hey, I want to do an internship but not in the summer. How about on demand?

    So [CHUCKLES] we've built basically like an internship as a service thing that projects can do. And so it's been a fun thing because LF, now we're over 200 employees-- probably 40 to 50 of these sub foundations. And we're constantly looking at ways to improve these developer communities to provide more services outside of just being a code host out there.

    ADAM GLICK: One of the neat things about your role is you get to see all the projects that are being proposed. You have a very good view of what's going on across the Cloud Native community and also a good sense for where things are going. What are your predictions for where you see Cloud Native and the CNCF going in 2021?

    CHRIS ANISZCZYK: I'll break those apart into two things. There's Cloud Native and CNCF. Given that we talked a little bit about my Eclipse IDE experience and so on, I'm a huge proponent of just improving the developer experience overall for Cloud Native and Kubernetes developers. I think finally now that A, we've agreed to a general orchestrator in the industry-- people have embraced YAML, whether they love or hate it-- we're moving to a point where IDEs are finally modernizing, where if you looked at, for example, the VS Code plugin for Kubernetes, it's gotten a lot better. It's pretty nice.

    But what other trend you're seeing now is I don't know if you've played with GitHub code spaces or Gitpod, for example, but essentially this whole idea is dev environment as a service. So basically, you could specify in your YAML file what you actually need to properly build everything and get a development environment set up. So if a contributor comes to your project, they could literally just click, Open in Gitpod or open up in GitHub Code Spaces--

    CRAIG BOX: Or Google Cloud Shell.

    CHRIS ANISZCZYK: --or Google Cloud Shell, yeah, there's lots of options. I haven't played with that one yet. But behind the scenes, it would go fetch the docker container image that you specified, set everything up, and open up VS Code style experience, where you contribute, debug, and build everything. To me that's where things are going to go, given that the technology there, I think is finally ready to enable that. So I think you're going to see a lot of innovation in what I call the Cloud Native IDE space.

    That's great. If you go to, for example, Prometheus-- one of our projects-- uses Gitpod. If you go to the Prometheus website, or GitHub repo, there's a little badge that says open up in Gitpod, and, boom, everything a developer needs to do to actually contribute to that project is booted up in 30 seconds. That's awesome. Back in the day, that's been a nightmare. I tried to build a business around standardizing people's Eclipse development environments back in the day because, what happens on your first day at work?

    CRAIG BOX: You compile your Gentoo installation. Then you install the JVM.

    CHRIS ANISZCZYK: [CHUCKLES] Yeah. You get all that configured, all your editor settings, you argue about tabs or spaces, depending which language you're using, and having that just configured and in a shareable construct-- so you could share with other developers-- is huge.

    ADAM GLICK: It's dev_setup.bat as a service.

    CHRIS ANISZCZYK: [LAUGHS] Yes, We wanna go that way. So I think you're going to see a lot of excitement in this space moving forward. A couple other areas we're seeing a lot of interest, at least since I could see the Sandbox project backlog that we have in terms of applications in CNCF, is there's a lot of WebAssembly, or Wasm, related projects that are starting to pop up.

    In the CNCF ecosystem itself, there's a couple of trends. One, for folks that were using either a JavaScript or a Lua-style VM for scripting or whatever, are seeming to move to Wasm for this purpose. So Envoy is a good example where they have this built-in Lua mechanism where you could write filters and Lua. That seems to be transitioning to be done in Wasm in the future.

    There's other projects there that are trying to figure out how do I run Wasm in a Kubernetes context. So there's folks-- I think Microsoft put together something called Krustlet. There's a couple of other Wasm container run times out there that are trying to accomplish this task. So I think you'll see more Wasm and Cloud Native in the coming year, even though Wasm is still very early days, and they have to improve the spec to support everything you need in running Wasm outside the browser. There's the WASI spec, which is improving.

    CRAIG BOX: Do you think that this is a case of you'll support running Wasm workloads instead of containers? Or do you think this is a use case where you will have an orchestrator that does to Wasm workloads what Kubernetes did to containers?

    CHRIS ANISZCZYK: I think the latter there. If you look now, we have projects like KubeVirt that treat VMs-- orchestrated like we would orchestrate containers in Kubernetes. I think Wasm will just be another workload, side-by-side, potentially, to these, that the Kubernetes ecosystem supports. On top of that earlier case, where I mentioned where folks like Envoy may be dependent on something like Lua, they're just going to rip that apart and use Wasm to handle their little extension mechanism.

    CRAIG BOX: Being able to run lighter weight workloads opens up low edge use cases for lower-powered devices as well. There's a lot of new projects in the space. Is this something you see as being a direction of the Foundation?

    CHRIS ANISZCZYK: Yeah, absolutely. Outside of late last year, we had the K3s project [CHUCKLES] join the CNCF sandbox. I have the unique position to talk to a lot of our member companies from all different industries. And it just seems like the last, I would say 6 to 12 months, the telco industry has just woken up and all of a sudden is very much set on making Kubernetes the dominant platform to run edge-based workloads in their 5G in future settings. So you're just going to see a lot of interest in stretching Kubernetes to support those use cases.

    And the way we talk about it internally is it's kind of just like Linux. Linux started out-- Linus posted something on usenet, a very hobbyist type thing. And eventually, end users took Linux, stretched it to work on embedded platforms, real time settings, mobile-- obviously, Android took over the world in many--

    CRAIG BOX: ...non-386 CPUs.

    CHRIS ANISZCZYK: I think you're going to see the community stretch Kubernetes in that setting. And it's going to be interesting to see how that is done, because I feel the maintainers have a stressful duty to ensure that Kubernetes works and doesn't break for anyone. So how these kinds of things happen over time is going to be interesting to watch.

    But my feeling is this year is going to be a lot of people stretching Kubernetes to make it useful on the edge. And we will be doing our first KubeCon Europe. We're going to be doing a Kubernetes Edge co-lo event in May. I'm excited to help foster that adoption.

    ADAM GLICK: Finally, you're fairly active on Twitter. You clearly were there fairly early on, as you have a three-letter alias.

    CRAIG BOX: And a Twitter mug.

    CHRIS ANISZCZYK: Oh yeah, I do have a Twitter mug [CHUCKLES] right now.

    ADAM GLICK: Three-letter acronyms are often used by government agencies-- NHS, IRS-- have you run into any naming conflicts?

    CHRIS ANISZCZYK: Absolutely. This is something I didn't learn because like I mentioned earlier, ZX Spectrum was my first thing. And so the handle I've been using forever has been ZX. I couldn't get that at Twitter at the time. There was an inactive account policy, where if someone-- if they didn't log in for a year, an employee could request the handle. And it was 11 months in, and this guy in Japan owns ZX, and he logged in, and I was like, no, I can't get it.

    [LAUGHTER]

    And so I was so angry. And he didn't tweet at all, which drove me crazy. He just logged in to read. Most people just log on to Twitter and read something. Oh no, he doesn't even tweet anything.

    So I'm like, I'm going to use my initials. And so I was like, oh, let's get-- CRA's available. Eventually what I learned throughout the year that is also the initials for the Canadian version of the IRS, the Canadian Revenue Agency. So--

    [LAUGHTER]

    --throughout the year, I get these angry tweets from Canadians complaining about the CRA. And sometimes I have to correct them like, complain to the Can Rev Agency. But I'm like, “look, Canadians, compared to Americans, you get great health care, great social benefits, stop complaining, and pay your taxes”. So that's my kinda annual reminder for folks. But yeah, downsides of short Twitter handles.

    CRAIG BOX: I guess it reminds you every year when the time for tax filing comes up in Canada?

    CHRIS ANISZCZYK: Absolutely. I have to mute people all the time. It's just like, you know what, stop complaining. You have great benefits, great health care [CHUCKLES].

    CRAIG BOX: Chris, thank you very much for joining us today.

    CHRIS ANISZCZYK: No worries. Great to be on.

    CRAIG BOX: You can find Chris Aniszczyk on Twitter @CRA, or on the web at aniszczyk.org; and you can find out how to spell that in the show notes.

    [MUSIC PLAYING]

    CRAIG BOX: Thank you for listening. As always, if you've enjoyed the show, please help spread the word and tell a friend. If you have any feedback for us, you can find us on Twitter @kubernetespod or reach me, at least, by email at kubernetespodcast@google.com.

    ADAM GLICK: You can also check out our website at kubernetespodcast.com, where you'll find transcripts and show notes, as well as links to subscribe.

    CRAIG BOX: I'll be back next week.

    ADAM GLICK: And I hope to see you all around the community. Take care.

    [MUSIC PLAYING]

View More Episodes cloud_download share