ABDEL SGHIOUAR: Hi, and welcome to the Kubernetes Podcast from Google. I'm your host, Abdel Sghiouar.

MOFI RAHMAN: And I'm Mofi Rahman.

[MUSIC PLAYING]

ABDEL SGHIOUAR: This is our KubeCon North America 2024 coverage episode. Kaslin spoke to attendees on the show floor, asked them questions about their experience of the event and some behind the scenes.

MOFI RAHMAN: But first, let's get to the news.

[MUSIC PLAYING]

ABDEL SGHIOUAR: Search Manager is a CNCF graduate project. Search Manager is a popular certificate manager for internet applications, allowing developers to automate TLS and MTLS certificates issuance and renewal. Congratulations Search Manager for achieving this milestone.

MOFI RAHMAN: Dapper also joined the list of graduated CNCF projects. Dapper, or distributed applications runtime, is a portable runtime that provides integrated APIs for communication, state, and workflow for building production-ready applications. Congratulations Dapper for this achievement.

ABDEL SGHIOUAR: Istio released version 1.24 and with that, announced that issue ambient mesh is GA. The Z tunnel and waypoint proxies, which constitute the core features of ambient mesh, have been marked as stable and ready for production-use cases.

MOFI RAHMAN: At KubeCon North America 2024, the CNCF announced the cloud-native heroes challenge, a bounty program aiming at helping fight patent trolls. During the keynote of the event, patent trolls have been raised as an issue Kubernetes and other cloud-native projects have to deal with. The bounty program asks developers to find pre-existing public information showing a pattent should not have been granted in the first place.

ABDEL SGHIOUAR: 2025 is fast approaching and the cloud-native community is already gearing up for a plethora of events. The CNCF announced the lineup for their flagship events for next year. On the menu, five KubeCon and cloud-native events in Europe, China, Japan, India and North America. One open-source security con in North America and 30 Kubernetes community days around the world.

MOFI RAHMAN: Three new cloud-native certifications have been announced at KubeCon North America, Certified Backstage Associate, OpenTelemetry Certified Associate, and Kawano Certified Associate.

ABDEL SGHIOUAR: Speaking of certification, the Linux Foundation announced that prices will go up by 10% starting next year for the three main Kubernetes certifications-- CKA, CKS, and CKAD, and the Linux certified system administrator exams.

MOFI RAHMAN: Wasm Cloud joined the CNCF as an incubating project. Wasm Cloud is a project built on top of reusable WebAssembly components, allowing teams to run polyglot applications on Kubernetes, Cloud, and Edge.

ABDEL SGHIOUAR: Spectro Cloud announced in a blog post that they have raised $75 million in series C funding. The San Jose-based startup intends to use the funds to help develop their Kubernetes management solution for on-prem Cloud and Edge installations.

MOFI RAHMAN: Solo announced they will donate their glu API gateway to the CNCF. Glu is an API gateway that runs natively on Kubernetes and enables developers to manage API endpoints and ingress access on their clusters.

ABDEL SGHIOUAR: And that's the news.

[MUSIC PLAYING]

KASLIN FIELDS: Hello, and welcome to the Kubernetes Podcast from Google. I am at KubeCon CloudNativeCon, North America 2024 and I am speaking with--

RAJAS KAKODKAR: I'm Rajas. I work at Broadcom doing all things Kubernetes there, and I also am a tech lead at Tag Runtime and heavily involved in the CNCF working group for artificial intelligence. So that's where I hang out these days.

JEREMY RICKARD: Jeremy Rickard from Microsoft.

RAY LEJANO: Ray Lejano, and I'm a solutions architect at Red Hat. I'm also a Kubernetes SIG Docs co-chair and a Kubernetes SIG security subproject lead, currently leading the external audit sub project. And I've also been a release lead as well for the 123 release team.

JIMMY ZELINSKIE: Jimmy Zelinskie-- I'm the co-founder and chief product officer of AuthZed. We build a product called Spice DB, and Spice DB is a database that is specialized for storing authorization data. I actually gave a talk on it briefly earlier at the conference, and largely, it was about securing Kubernetes using spice DB because Kubernetes authorization is a little bit work in progress, if I would say so myself.

FREDERIC BRANCZYK: Frederick Branczyk. I'm the CEO and founder of Polar Signals. We build tooling to measure and analyze performance over time. So that way you can lower resource costs across your cloud bill and improve performance, but also down to the source code line number, understand why your code is behaving the way that it is in a very unopinionated way because it requires no instrumentation. So it's truly just what the source code is. If the source code is executing, that's what you'll see.

LUCY SWEET: My name is Lucy. I'm a senior software engineer at Uber, and I work on our Kubernetes and compute platform.

SREEKARAN SRINATH: Hi, I am Sree. I am a full-time software engineer and part-time grad student.

JOE THOMPSON: Joe Thompson-- I'm a consulting engineer at clarity business solutions based in Linthicum, Maryland, and I've been part of the Kubernetes community for almost 10 years now.

KASLIN FIELDS: What were you hoping to get out of the event?

RAJAS KAKODKAR: Mainly, I was trying to look at how we can integrate AI with Cloud Native. Like, that was like a hot topic this time. There has been a lot of buzz around scheduling workloads. It's almost impossible to go to a talk right now without people mentioning queue.

And I was just trying to flip things around and see what's on the other side, like take an upside down approach and see what all we can do from learning from AI and integrating that with our ecosystem. And it's been pretty good. There have been like a couple of fun thoughts around that, things around issues with GPU monitoring, extending Kubernetes in a way that it adds to more resource scheduling or more efficiency in terms of how we do things. So that was really good.

JEREMY RICKARD: I was really hoping to connect with contributors that I haven't seen in a while. I was really excited to have a lot of face-to-face discussions at the Contributor Summit, and that worked out really well. I think we had a couple of really good conversations about LTS.

I'm a co-chair for the working group LTS, and I think having those meetings bi-weekly, the regular cadence, is great, but in-person is so much better. And we had really good focus time to work through some challenges and some issues that we think we're facing. I'm still seeing a lot of security related things. I think even at Rejects, there was a lot of security content and it was really interesting to see.

I did a talk on security as well, and I had a lot of conversations with people afterwards. And it seems like there's so many common problems that people are facing right now reacting to ever-growing security, vulnerabilities, and just the complexity of managing that. And seeing some of the stuff in the vendor space, the tools that are coming out for that is really cool, and seeing what projects are doing to increase security.

I think also the thing that has been really cool to see is the momentum around SDO. So I took away the news about ambient going GA. And also, in the keynotes yesterday, there were some really cool things about envoy-based app gateways that I thought was cool.

RAY LEJANO: For KubeCon, it's always about connections. So I love connecting with my fellow contributors at the Kubernetes Contributor Summit, which is the last one, actually. It's going to be the maintainer summit at the next KubeCon.

I also love meeting new contributors as well at the SIG meet and greets who want to get involved with contributing to Kubernetes. Also just meeting new to me Kubernetes contributors and just talking about their projects or what they're working on.

JIMMY ZELINSKIE: Yeah, I think this conference, specifically, is mostly my talk. My talk was supposed to be kind of enlightening folks to kind of the inadequacies in Kubernetes authorization system. Basically, get people up to date on the history of authorization, let them understand what exists in Kubernetes, and then kind of where it falls over, and then kind of where Kubernetes is going in the future and then how we're going to have to adjust that foundation to support that future.

FREDERIC BRANCZYK: So I think being naturally drawn to low latency, high performance workloads, we wanted to understand even more what kind of workloads are out there, because we're always surprised just how many problems in the world are latency sensitive, performance sensitive. And we just had Jimmy on as well that there are definitely also a super low latency workload. And there are things like this that we find new things every time we come here. So we're always excited to come back.

LUCY SWEET: It's always important to us to come to these things, to meet up with the contributors, and to see what's next and what we might be missing in our platform. I've seen a lot of great talks. I saw some great new things around stopping disruption of stateful workloads. And it's always important for me to go to the Contributor Summit to see our contributor family and to work out what we're building next in Kubernetes. We've got some really great stuff coming in SIG node that I'm really excited for over the next few releases. It was great to align everyone at the summit there. And I'm looking forward to it coming out over the next few years.

SREEKARAN SRINATH: I was hoping to get a 101 on Wasm, which I did. I was hoping to make new friends and catch up with old ones. It's been a couple of years since I last came to KubeCon. And so I wanted to get the usual six month boost where I am very excited about contributing, getting back into things. It usually lasts till the next KubeCon. So yeah, I was looking to get that out of this, and I think I did. So yeah, time will tell.

JOE THOMPSON: For this year, I was hoping to reconnect with some more people that during the pandemic and we weren't doing things in person. There were some in-person meetings that I missed. Not everybody was back at KubeCon in person last year. It seems like more people are back in person this year. I ran into a lot more people in the hall and side events and that was fantastic.

KASLIN FIELDS: What trends have you seen at the event?

RAJAS KAKODKAR: Particularly AI.

[LAUGHTER]

But the one main trend that I'm seeing is the focus on security and how it's not just about running the workload, but making sure that the workload is hardened. The entire journey of a particular workload is done in certain ways so that there is track or at a station at every stage of it. So that's like a good focus that I think we should be having for this decade going forward. And I'm seeing a lot of involvement from the end users as well, a lot of case studies. So that was really good. Yeah.

RAY LEJANO: So I think the big trends are actually aligns with the themes of the KubeCon keynote. So day one was AI. Day two is security. And day three is community. And also the interrelations between some of those themes, like with AI, with how security is applied to AI, with maybe bombs for LLMs, and also like security applied to gateways as well.

JIMMY ZELINSKIE: Yeah, trend-wise, I think there's kind of two things that come to mind. The first one is I actually got put on the keynote in Paris for my opinion on this, and I think it's actually coming to fruition. Heterogeneous workloads on Kubernetes. So there's a lot of talk about running different types of workloads.

We talk a lot about running low latency workloads on Kubernetes. I know LinkedIn just made a big post about how they do stateful services on Kubernetes, slightly different from stateful sets built-in. And then there's obviously tons of talks on AI workloads.

So that is the more interesting evolution of even container workloads. But then outside of container workloads, there is a lot of traction around how you make Kubernetes kind of core APIs extendable beyond just the container workloads. So you see projects like KCP kind of testing the waters there for a future of Kubernetes that's kind of more extensible than what we have today, and less container focused.

FREDERIC BRANCZYK: So again, this is a little bit related to the previous question. I think what's blatantly obvious is everything AI that's happening, but there's sort of an extension that we're seeing more and more of, which is there are many more robotics companies popping up. And specifically for the work that we do with low latency and high performance workloads, this is even more important when we work with AI in the physical world. That's a trend that we've been observing very recently.

LUCY SWEET: I think a big thing that I'm seeing, particularly around AI and generative workloads, is that in the past, we used to have that everyone was on the same hardware and the same arches, and no one really cared about what hardware they were on in the same way, or at least not many people did. Back in 2014, 2016, you were either on Intel x86 or AMD 64. Kubernetes itself was built around this time, and you can tell because we just have CPU and memory. It doesn't really matter which memory.

But more and more I'm seeing that users and developers really do care about what hardware their workloads are using. Obviously, the great example and the obvious example is GPUs. Where they're like, I need to use a H100 because it's better at training. But even outside of that, a lot of people are moving on to ARM because it's cheaper and it's more energy efficient for them.

And Kubernetes is having to respond to that. So we've got-- and we're responding quite well, I think. We've got DRA coming down the pipeline, for example, and that has been a great boon for this. But I think that this is becoming more of a trend in our industry, that we are becoming more fragmented and more specialized in the hardware we're using. And I don't think this is going to stop in the near future.

SREEKARAN SRINATH: I think this one's a little obvious. AI and more AI, we saw that coming, but it's a bandwagon. But I'm curious to see where it goes.

JOE THOMPSON: AI and LLMs and machine learning, and I distinguish between all three of those, continues to be big, just as it was last year. Machine learning was big a few years before that. I think one of the trends that I'm starting to see now is the idea of governance of LLMs, not only in terms of the ethics of things, like training and the source material, but also things like how much resources they use and how are we going to run those things in a shared environment with other things without everybody stomping on each other. The other big trend I'm seeing is in Kubernetes itself, there seems to be a lot more talk this year about observability.

KASLIN FIELDS: What's your favorite thing you've learned so far?

RAJAS KAKODKAR: One thing that came out of this conference was the focus on quantum machine learning for Kubernetes. I've just been hearing about all of these things. And yeah, I really like-- I don't know what it means. I don't know what I'm talking about over here, but I've been hearing this from a bunch of folks using quantum machine learning, and that sounds pretty cool. So yeah, that's really something that I learned.

JEREMY RICKARD: I think the thing that I found the coolest was from the Contributor Summit, there was a session talking about the work that's happening for API compatibility. Again, going back to my working group, LTS, with that hat on, I think one of the questions or one of the problems that we saw through some of that work was that people are really hesitant to upgrade because they're afraid that there will be changes in the API that will break their workloads.

So that's one of the hesitations they have with upgrading. And compatibility mode, I can't remember the cap number, but introduces a new flag to the API server and eventually would bubble down to the rest of the components that would allow you to run with a targeted API version.

So you may be able to use a binary that's 132, but set it to Kubernetes 130 as the compatibility version, and it would really only expose the things that were in that version. So it would give you a little bit more protection or maybe a little bit more comfort or warm fuzzies about upgrading, and then you can bump it to the later version after you've worked through some of those challenges.

RAY LEJANO: So far, I met a fellow Kubernetes contributor who works on the COSI, or Container Object Storage Interface, which allows you to provision or to consume object storage in Kubernetes. And it's also an alpha feature. And so we had a great discussion on how to make that a COSI more secure.

JIMMY ZELINSKIE: I think the thing that I learned the most is actually not at all coupled to Kubernetes. There was a talk called the Maintainer Monologues, and in it, it was mostly about different open source maintainers talking about their kind of stories. And there is really great advice from Scott Rigby about avoiding burnout by actually building concrete workflows so that when someone is kind of crossing a boundary, you have an established pattern and policy for handling someone crossing the boundary, and that will keep you and others safe from people taking advantage of you in ways that aren't sustainable long-term.

FREDERIC BRANCZYK: I think this is, again, a little bit related to AI, I think. We've been learning more about the hype seems to be dying down and we're kind of coming to the reality of how do we actually operate these AI workloads. And I think there's still a lot of challenges to be figured out here. And I think this is exciting, right? It would be kind of boring if it was the same old stuff all over again. So it's exciting to see that there are complicated and interesting problems that are being tackled by the community.

LUCY SWEET: I went to a talk where someone was demoing a really cool thing that they'd built where they were using Firecracker and a few changes to the CRI in order to run an application and move and reschedule it on a different node without stopping the application. And I found that really, really cool. And it has, in my opinion, great use cases for things like databases and things where, guess what, you don't have to reason about disruption as much anymore.

I think it's probably like a year or two away from being like production ready. It seems what they had done was quite hacky, involved quite a lot of forking of like storage, for example. But I think that stuff like this and is going to be especially useful, especially around reducing disruption. And I mean, we see a similar trend in Signode where we're working more and more on less disruption when you do stuff to your pods.

SREEKARAN SRINATH: Well, like I mentioned, the Wasm 101, I had no clue what it was coming in, but I've heard my friends talk about it a lot. And WasmCon gave me a good opportunity to get started. I think I'm going to look into it a little bit more.

JOE THOMPSON: Favorite thing I've learned so far, I was pleasantly surprised to learn that the in place pod resize is actually going into beta in 1.32. I was actually talking to somebody yesterday who had thought that it was in 1.31, and I was like, nope, nope, it's in 1.32. So yeah, eagerly awaiting that. The 1.32 betas are out. Come on, 1.32.

KASLIN FIELDS: What would you like to see at future KubeCons?

RAJAS KAKODKAR: One thing is that, I mean, we should keep this momentum going because I truly like the energy over here. And specifically I would like to see more talks or panels from the academia, the research side of things, and more case studies coming out from academia. The poster sessions have been like a really good step and more case studies would be really good.

JEREMY RICKARD: One of my favorite things is seeing end users, and I think I'd love to see more end users. I think I've been asked that question before and it's probably the answer I'll give in the future. Just, I love seeing how people take the things that we're working on and deliver cool stuff.

RAY LEJANO: For me, at KubeCon, it's not just learning new things. It's also about growing the community as well. So I think when my personal passion is tutorials, I've done free tutorials at KubeCon, so I'm more hands-on workshops for new contributors or just folks just to engage with the community, with maybe with other CNCF projects as well.

JIMMY ZELINSKIE: Yeah, I think KubeCon is always kind of a work in progress. So it's nice to see the subtle changes they make every year. Honestly, I think the Project Pavilion being continued to be invested in is definitely like a big one. I know we've done a lot of experimentation, so trying to pull people in there. There's tours this year and they kind of have the big stage over there to bring people in. I just think more attention needs to be paid there.

And then also this year, they kind of did a weird thing where they kind of closed the sponsor hall while talks were still going and a bunch of people ended up leaving and not actually even knowing that talks were still going on. So maybe better communication about that or at least helping guide people towards those rooms near the end of the day.

FREDERIC BRANCZYK: So I don't know if I'm necessarily representative of the larger KubeCon crowd. And I think this is something that we've been trying to address with the pre-events, but I would love to see more in-depth technical talks again at the main event as well. And I feel like I keep hearing this from the larger audience as well.

I do think we've done a good job of that at the pre-events, but I think we can allow some more of that back into KubeCon proper again as well. So I think that's something I'd love to see. And I feel like the CNCF projects maybe could get a little bit more attention as well at the conference. But overall, I think we're moving in the right direction with that.

LUCY SWEET: I think it's always great to be here, and it's always a great conference. But I think one of the things that I wish I saw more of was stories of end users at KubeCon. There's a lot of vendors, there's a lot of integrators, and that's great. But sometimes I just want to hear, like, implementation stories and war stories, and to really try and get a sense of have I done something that someone else has done in a better way than me, especially as coming from a company that is an end user, it would be great to see more end users here.

SREEKARAN SRINATH: I definitely want more technical content, more workshops, stuff where I can get my hands dirty. I did enjoy the Contribfests. I would definitely like to see more of those and a little better CFP filtering. A lot of the talks seemed promising when you looked at the titles and the abstracts, but fell short of expectations. Yeah.

JOE THOMPSON: Really more of everything. I'm hoping that KubeCon will get back to the pre-pandemic numbers in fairly short order. That would be nice. And it would be nice to be able to do that safely. One thing that we need to do more of, it seems like they've backed off a little bit from the virtual component of the event, and I would like to see that return in a more-- because that's a big accessibility thing for a lot of people who aren't able to attend, for whatever reason they're not able to attend. So more emphasis on the virtual part of the event would be nice.

KASLIN FIELDS: Wonderful. Thank you very much.

[MUSIC PLAYING]

ABDEL SGHIOUAR: Well, hey, Mofi.

MOFI RAHMAN: Hey, Abdel.

ABDEL SGHIOUAR: Unfortunately, Kaslin could not join us this round for logistics reasons. So people have to deal with me and you, I guess.

MOFI RAHMAN: I mean, you make it sound like it's a bad thing. I don't know.

ABDEL SGHIOUAR: I don't know. I think people will have to tell us.

[LAUGHTER]

Yeah, anyway, so if you are listening to this, we are the week after KubeCon. This, you have listened to the KubeCon episode, obviously. We were all there in Salt Lake City. Off the top of my head, one of the things I've heard a lot is a lot of people were surprised how lovely Salt Lake City was. I don't know. What was your impression?

MOFI RAHMAN: No, I think it was a lovely little place. Every direction there is a mountain, which I was not expecting. I did not, like, from the location, pretty much all direction you look, you see a mountain range. We got a couple of small days of snow as well, which was kind of like good and bad for things. But it was like the snow on top of the mountain looked lovely.

ABDEL SGHIOUAR: Yes. Yes. But the city was nice. The blocks are huge. A lot of walking, but it was cool. But anyway, KubeCon, how was Kubecon for you?

MOFI RAHMAN: I mean, like every KubeCon, KubeCon is always part work, part talking to people, but part also meeting co-workers all the time, and friends and other people in the community. So I feel like even the worse KubeCon is generally pretty good.

ABDEL SGHIOUAR: Yes.

MOFI RAHMAN: So it's hard to have a bad KubeCon generally. But this KubeCon especially was really good. Kubernetes turned 10-years-old. So you have in every KubeCon, we still get about, what, 40% to 50% people that are coming to the KubeCon for the first time.

A good percent of them using Kubernetes for the first time in their life in that year alone. So as the community is growing, we have a lot of new people joining the community as well as coming to KubeCon. So that's always exciting to see.

ABDEL SGHIOUAR: Yeah, it's still, like, that's one of the most interesting stats that comes out of every KubeCon is the number of new joiners. We covered this in the news, but I think what was particular or what was interesting about this KubeCon is that the opening keynote was talking about patent trolls. There was like a good 10 minutes talking about that.

MOFI RAHMAN: I mean, that's kind of the sign of a project I feel like getting more mature. Like, Kubernetes over 10-years-old. In the first five years of Kubernetes, patent trolls probably wouldn't have cared about Kubernetes enough to, right, do trolls to. Have patent trolls in your particular line of technology in some convoluted ways a sign of how popular the technology has gotten?

ABDEL SGHIOUAR: A sign that you are doing a good thing?

MOFI RAHMAN: A sign you are doing a good enough thing for someone to seek illegal monetary benefit from it, right? So it's--

ABDEL SGHIOUAR: Well, it's legal. Let's say unethical. It's legal for now.

MOFI RAHMAN: Yeah, that's the better word. Yeah, it's not illegal. It's unethical. It's kind of-- yeah, I think unethical is probably the better word. Not illegal.

ABDEL SGHIOUAR: Yeah. Anyway, so what was your takeaway there? Anything interesting that you saw, heard, talked about, chatted about?

MOFI RAHMAN: Like AI and other things that are top of mind for a lot of customers and teams figuring out how to make Kubernetes a better place for it. I think the workgroup serving got announced. I think last year, kind of this KubeCon, but I think the group got bigger this cycle. A lot of people are talking about serving as a dedicated type of workload on Kubernetes. For the longest time I probably was saying like, serving is just for application.

That's another different type of web deployment. But after spending some more time into that space, I probably do agree that there are some nuances that require a little bit of extra attention. Maybe from the Kubernetes side of things or from the provider side, just the need for additional resources, like GPUs and TPUs, make those workloads a little bit more unique compared to your traditional web application.

ABDEL SGHIOUAR: Yes, and we did have a WG, a working group serving episode a few episodes ago. We're going to leave a link in the show notes. But that definitely was-- I mean, we learned that when we were preparing for our sessions. There's quite a lot of challenges related to how fast your workload starts because you're typically downloading like a very large model and doing multi-host serving, sharing GPUs. These are all things that are still kind of in the working basically. Autoscaling based on serving workloads is also another one. So definitely a lot of that. Yeah.

MOFI RAHMAN: Yeah. So like from that autoscaling point, right, like, so your traditional HPA way of looking at CPU and memory metrics is not necessarily the best way to scale for your AI workload. That requires looking at GPU metric. So there will be work in the next few cycles to figure out, how do you either improve HPA or plug into something external, like KEDA, in your application system to scale, not just based on your CPU and memory, but something else that gives you a better signal of how your application, your inference serving application are doing on your cluster.

ABDEL SGHIOUAR: Yeah, we did record an episode on the YouTube channel about autoscaling based on TGI size, queue size. So if you use TGI, which is the Hugging Face Serving library. It does have a metric called queue size. So basically, based on how much requests are coming from users, the queue is essentially the amount of prompts that are waiting to be processed by the LLM. So there is definitely some movement in that space. As you said, it's quite an interesting-- it's quite interesting space to keep an eye on.

Yeah. So, I think for me, some of the interesting things, I mean, Istio had a big announcement to make, and Solo, the company behind it, had big, big presence at the event. So that was quite interesting because they were just right behind our booth. There was quite a lot of other big vendors present with big announcements that have been made at the conference. But I feel like this edition, a lot of the big announcements have been the CNCF itself.

Like, the certifications, the price, all these new things coming out of the graduated projects. Obviously, it's the end of the year. So they had to celebrate the new graduated projects. So that have been to me, the most interesting thing going on is like all the stuff that the CNCF announced. I don't know if it's all KubeCon North America because this was my first time going to KubeCon North America, but it was definitely like something that was kind of unique and different from the European version.

MOFI RAHMAN: No, I think the announcements are always great. I think this year another kudos to CNCF for highlighting the Project Pavilion in a better way. I think they had this big group of people to project, to project, to get people to talk about all these different projects.

Project Pavilion also was situated, I think, more of a central location where people could easily find them. So that was, I think, a big win for people to be able to find this open source project that a lot of people rely on.

So it was good for people to know about different projects that they probably otherwise wouldn't have heard about. Also, for the projects to be able to recruit new contributors, and maintainers, and members. So I think that was a big win for me. I went by the Project Pavilion multiple times just to talk to people about different projects. And yeah, that was exciting stuff.

ABDEL SGHIOUAR: Yeah, and we spent quite a lot of time also recording some content, which should be coming out soon. Video content to be more precise.

MOFI RAHMAN: Yeah, we got to interview a few different projects and few people. Obviously, there were a lot more than we could have covered, but it was a start. Hopefully in the future we'll get to do more of those.

ABDEL SGHIOUAR: Yeah, and something I learned is that you used to do some photography. You taught me a lot about videography and photography.

MOFI RAHMAN: Yeah, it's kind of like, in a past life, I had a small business back in college. I actually used to do that. And back then, I was like, this skill will never come in useful ever again. But look at me now. It's coming useful.

ABDEL SGHIOUAR: Yes. Now you have cameras and stuff.

MOFI RAHMAN: Yeah, this is, I guess, a lesson to everybody listening. Any skill you have, if not directly, sometimes indirectly, can be used for other things.

ABDEL SGHIOUAR: Yes, yes. Definitely. Definitely. Well, cool. Thank you for joining me. Thank you for filling up for Kaslin. Although I don't think any of us could actually fill up with Kaslin, anyway.

MOFI RAHMAN: No. Yeah, so I think in Seattle, there has been a storm, as of we are recording this. So Kaslin is OK, but internet connection has been a little less than ideal. So we're hoping that gets fixed very soon.

ABDEL SGHIOUAR: Yes.

MOFI RAHMAN: And yeah, and anybody else that are listening are now, again, you're going to be listening to this in the future. But if you are struggling with the internet connection and the storm, our best wishes. And yeah, so that everything gets fixed very quickly.

ABDEL SGHIOUAR: Yes. We wish you the best. All right. Thank you very much, Mofi

[MUSIC PLAYING]

KASLIN FIELDS: That brings us to the end of another episode. If you enjoyed the show, please help us spread the word and tell a friend. If you have any feedback for us, you can find us on social media at KubernetesPod or reach us by email at, <KubernetesPodcast@google.com>.

You can also check out the website at kubernetespodcast.com, where you'll find transcripts, show nodes, and links to subscribe. Please consider rating us in your podcast player so we can help more people find and enjoy the show. Thanks for listening, and we'll see you next time.

[MUSIC PLAYING]