Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Kubecon North America 2023: Highlights, Themes and Key Takeaways
1.3k Plays11 months ago
In this episode, Ryan and Bhavin talk about Kubecon + Cloud Native Con North America 2023 in Chicago, and discuss all the vendor announcements from the past couple of weeks. Kubecon North America was in Chicago and had more than 13500 attendees and shows a continuous increase in adoption of Kubernetes as the platform to run containers, virtual machines, AI workloads, etc.
Check out the KubernetesBytes website: https://www.kubernetesbytes.com/
Ads:
- Ready to shop better hydration, use "kubernetesbytes" to save 20% off anything you order.
- Try Nom Nom today, go to https://trynom.com/kubernetesbytes and get 50% off your first order plus free shipping.
Find all the links to the things we discussed during the episode here: https://docs.google.com/document/d/169mrZGi7oYptuT1nXrmuy-uwIlrAYZ8jJR3p5vRZ5Vo/edit?usp=sharing
Recommended
Transcript
Introduction to 'Kubernetes Bites'
00:00:03
Speaker
You are listening to Kubernetes Bites, a podcast bringing you the latest from the world of cloud native data management. My name is Ryan Walner and I'm joined by Bob and Shaw coming to you from Boston, Massachusetts.
Podcast Agenda Overview
00:00:14
Speaker
We'll be sharing our thoughts on recent cloud native news and talking to industry experts about their experiences and challenges managing the wealth of data in today's cloud native ecosystem. We'll be right back after this short break.
00:00:32
Speaker
If you've ever had a puppy and raised it to become a big dog, you know that changing food and finding the right food is hard to get right. Ultimately, you want them to feel good and act happy and be okay with what they're eating. They're part of your family, after all. I have an eight-year-old golden retriever named Roscoe, and he's always had a sensitive stomach, so finding the right food is kind of a pain. That's where Nom Nom comes in.
00:00:57
Speaker
Nom Nom's food is full of fresh protein that your dog loves, and the vitamins and nutrients they need to thrive. You can actually see proteins and vegetables like beef, chicken, pork, peas, carrots, kale, and more in the ingredients.
00:01:12
Speaker
So here's how it works. You tell them about your puppy, the age, breed, weight, allergies, protein preferences, chicken, pork, beef, and they'll tailor a specific amount of individually packaged Nom Nom meals and send them straight to
Nom Promotion
00:01:25
Speaker
you. If you're ready to make the switch to fresh, order Nom Nom today and go to https forward slash forward slash trinom.com slash Kubernetes Bites.
00:01:36
Speaker
and get your 50% off of your first order, plus free shipping. Plus, Nom Nom comes with a money-back guarantee. If your dog's tail isn't wagging within 30 days, Nom Nom will refund your first order. No fillers, no nonsense, just Nom Nom.
00:01:55
Speaker
And we're back. Good morning, good afternoon, and good evening wherever you are. We're coming to you from Boston, Massachusetts.
Hosts' Greetings and Date Announcement
00:02:02
Speaker
Today is November 21st, 2023. I hope everyone is doing well and staying safe. Bhavan, how have you been? I know you've been doing a lot of travel. Are you glad to be home? I am. I'm sure that.
Travel Fatigue and Conference Exhaustion
00:02:13
Speaker
I was excited for the travel.
00:02:16
Speaker
before the travel. And now I'm just glad that yeah, I'm glad it's done. Last Thursday, I think in minus one day, before my conference was done, I was like, shit, I want to go home. I want to go home. I can't have excitement to talk to people at the booth. Like even nothing to the people at the booth. But it was just like, I'm done. I didn't have the energy. Your battery runs out. Everybody's got a battery. Some people's run out faster than others. And it'll happen everyone. And I don't blame.
00:02:45
Speaker
I don't blame you for being like, I don't want to talk to a single person today.
Denver Travel Stories and Scenic Experiences
00:02:49
Speaker
But one cool thing, like when I was in Denver for the supercomputing conference last week, right, Tim was there to like, I know you Tim, you know Tim, but he drove me to his place in Estes Park and I met his dogs and like, man, they are so cool. Like they just hang out with you. And like they immediately like we started being your friends.
00:03:09
Speaker
Like, OK, I was super pumped to meet his family and the dogs. You got to see his mountain as well. Yeah, dude, it's a crazy driveway. I was like, I couldn't drive like with with with his truck. I was like, I don't know how you are making these turns with obviously has done that for like thousands of times, but still like it was a crazy, crazy and everything. Yeah. Yeah. Those stories. I'm I'm like jealous, but no, it's a lot of work all the same time. But it is a beautiful piece of property. I saw he posted that like
00:03:38
Speaker
KubeCon Recast for a video where he climbed to a nice view and stuff. I was like, to be able to do that in your backyard. That's his own property. He didn't have to go to a state park or a national park. He was just right there. I know. It's a really cool location, but again, as you said, a lot of work. I kept telling him on our way there, I can't do this. I can't do this to us. Dude, I'm barely able to keep up with the condo in Arlington.
00:04:06
Speaker
I'm not going to be able to manage and maintain and sustain a cabin in SS Park. Hey, to each their own, right? How was your KubeCon travel?
KubeCon Travel and Listener Feedback
00:04:19
Speaker
Yeah, no, I can't complain. It was pretty easy. I know I had some travel right beforehand. And so I got back from KubeCon feeling the same way like glad to be here. Although I think I came in from KubeCon 11 p.m. and Boston and then
00:04:38
Speaker
I went to bed when I got here, it was like an hour away, woke up packed and we drove to Vermont, so I wasn't going for a break. Okay, you still had travel, yeah. But that was like a family getaway, travel a little bit, just take some time off, which was really nice. That's awesome. It was up there for the weekend, and it was cool because the whole, we're near Stowe, Vermont, which is beautiful.
00:05:03
Speaker
The season when you're up there at that time of year, like right in the first few weeks of November, sort of like Vermont's half and half in winter and still in fall. Like half the tops of the mountains are all snow, but then the bottoms are green still. And then you can take pictures where like there's snow falling in the background, but you're still in like a green pasture and stuff like that. So it's kind of like this cool.
00:05:31
Speaker
Yeah, I didn't know you were in Stove because like last time we drove there, I think in September, we found like a really cool cocktail bar. Like it had like the setting of a ski resort, like an apree. I guess if I don't know if I'm saying that right. Yeah. Yeah. And it was really cool. Like good drinks. Yeah. It was very cool. It's got, I mean, for a Vermont town, you got to, you got to like put it in perspective. It's never going to be big if you're in any town in Vermont, right? They're like five restaurants. Yeah, exactly.
00:05:59
Speaker
There's yeah, there's there's a couple breweries a couple good restaurants, you know those kind of things and I feel like being Sort of remote. It's it's beautiful and we always like getting away there. So
00:06:11
Speaker
Yeah, that's what I did. And now we're back here and we're ready to do our KubeCon episode. I know, I'm always pumped to do these. Like I had somebody, like I know we talked about this about our KubeCon event, but I had some one of our listeners stop by the booth and like, what are you going to cover in the news? You have to wait. Nah, no sneak peeks. Yeah, we did have a few people come up and
00:06:34
Speaker
shake our hands, say hi. We really appreciate that because that gives living proof that people are listening, which is always really helpful to our own ears and egos. We're not just doing this talking to ourselves. Even though I like talking to Ryan, when we don't hit record, we're still doing the same things that we're doing right now. It's always good to get that validation. It's not just a bubble that we're talking in. There are people that are listening to us.
00:07:01
Speaker
Thank you. So thank you, thank you for being listeners and coming to say hi. That was very cool. Yeah, so we're post-CubeCon, pre-Thanksgiving, and we're going to do our news episode.
Focus on KubeCon News
00:07:14
Speaker
We like to do news in an extended fashion, but really around releases and things that came around that week, so to speak. So we also have some themes
00:07:29
Speaker
thoughts of our own that we'll bring into the fray here because you know our experience at KubeCon might have been a little different than some or may have been similar in ways so we like to share that and hopefully that's useful for folks who didn't get to go yeah as well so
00:07:44
Speaker
Let's start off with how the show was in general for you, Bhavan, and maybe how it compared to some of the other ones. Yeah, I think Cubecon Chicago kept getting better, right? I've only been to three Cubecons, LA, Detroit, and the one in Chicago. This one was the best for me. I think in terms of the number of people that I've attended, and for surprising, like I spent most of my time at the Portworx booth during the actual conference days.
00:08:08
Speaker
And there were a lot more people that were willing to have a conversation and ask questions. And I was trying to relate it back, right? Like, you remember the talk that we attended at DevOps Space Boston, where Pete was talking about zero interest rate projects or reserve projects, and how everybody wanted to spend developer time and solve things that were already solved in the vendor ecosystem. And I think that was happening till last year. Now, I think
00:08:35
Speaker
Things are getting expensive. You can't have developers just work on some smaller open source project that eventually helps you. I think the build versus buy discussion is happening inside bigger organizations. This is just my assumption. That has forced people to see what solutions are out there and how they can plug these into their existing stack rather than trying to reinvent the wheel because they want to
00:08:56
Speaker
make something for their own organization. So I think with that background, I had a lot of conversations around, again, for Portworx, right? How have things changed? What's new? How can Portworx help us? Things like that. But yeah, really engaging KubeCon for me. How about you, Ryan?
KubeCon Attendance Insights
00:09:12
Speaker
You know, I think, you know, when you look at Detroit, it was definitely bigger than Detroit. But if you look at the previous KubeCon in Amsterdam, it's actually smaller, right? So I think Amsterdam was around 10,000 and
00:09:25
Speaker
Gosh, Chicago's around 8,000. I think it's the numbers that I've been hearing. So it was less people. And I did see that in a few different ways, right? So like, I don't know if you remember, oh, no, you weren't Amsterdam, right? So yes, Amsterdam was bigger by a couple thousand people. And like the
00:09:44
Speaker
the cube, the sort of cube crawl, the nightly sort of thing that they have on Tuesday or Wednesday. That was definitely, it felt like a lot less people were in the show floor for that cube crawl on
00:10:01
Speaker
in Chicago. And I don't know if it was just the difference of a couple thousand people or if it was just the way the hall was set up or whatever, but you know, there was still a good amount of people there, but I noticed a little bit of difference and I'm not exactly sure why. And it could just be an EU versus America's thing in terms of interest or that kind of thing, but 10 to eight. And you know, I also had a couple of conversations around folks that
00:10:26
Speaker
Budgets are shrinking, unfortunately. So maybe there was less people that got the budget to fly and attend. But you know, it was a little smaller, but it still had a really good energy, like you said. Like I've heard different numbers. So I think
00:10:42
Speaker
We should see what CNCF eventually publishes. I've heard like 15,000 people at Chicago, not 8,000. I don't know. I mean, this was from the CNCF. Oh, it was? Okay. From someone who's involved. I think it was around eight. But 15 would be very surprising if that turned out to be the number. We'll confirm and put it in the show notes. But yeah, I think, you know, to your point where I think it did have a slightly different sort of feeling to it in the sense that
Kubernetes Maturity and AI Discussions
00:11:11
Speaker
I think it spoke to the maturity of where Kubernetes is, right? I read an article and I think I put it in our notes here about it was from Forbes, but the whole idea around it being that operators have arrived, right? It's not just developers who are part of the projects and project teams and open source community. It's like, no, we're sending people that have money and budgets and executives are now showing up to basically make decisions, right?
00:11:41
Speaker
as the way I took this article. And I wouldn't really disagree, right? I think people are getting to that point where it's like, okay, there's no turning back that this is a platform that's here to stay. So, you know, we're coming here with our checkbooks, so to speak. Not literally, I hope, but... Oh, I would love to see that, dude. Like cloud native, but the transactions are still done through paper. Yeah, that would be awesome. I'm sure it's been done before.
00:12:11
Speaker
Anyway, so I mean, what do you think? Maturity-wise, where are we at? No, I think from a majority-wise, I know, like CNCF mentioned, they've crossed the chasm a couple of years back. So we definitely see that evolution, right, in terms of the attendees. But even in terms of teams, like in terms of what everybody on the show floor or in the keynotes were talking about, it wasn't just about, oh, how cool Kubernetes is. It was always about like how people are using Kubernetes to do everything, right? And I know
00:12:40
Speaker
But the keynotes, AIM was like a huge thing. And we will talk more about AI because it was undoubtedly the favorite thing and term to talk about in Chicago. So it was always about how Kubernetes is that platform or that underlying layer that's supporting all different use cases. So agreed, right?
00:13:03
Speaker
We have matured enough that people are trusting Kubernetes as that platform or they are investing heavily into this and go and trying to find alternatives from whatever they were using today and see how Kubernetes fits into their use cases, fits into their deployments, and then selecting Kubernetes as that platform for the future. This can be to run containers, this can be to run virtual machines through the Kubernetes project that we have spoken on this episode. Things like Wasm, like serverless with Knative, you can run any kind of applications. Kubernetes is that data center OS that Ryan likes to talk about.
00:13:32
Speaker
I definitely agree that that was the trend. Kubernetes is here to stay. Yeah, absolutely. There's no ifs, ands, or buts about it.
00:13:44
Speaker
So, I mean, there's a whole bunch of other things I think, but before we dive into maybe some specific news, I did want to talk a little bit about
AI as a Key Theme at KubeCon
00:13:53
Speaker
the themes. We covered AI as an obvious theme. I think without a doubt was a standout theme, even though there was not like a ton of talks, like there was talks that had AI in it. No, I think my observation is AI was a theme for the keynote, but then when you actually walked around the show floor,
00:14:13
Speaker
it hasn't permeated enough that all the vendors are just talking about AI. Like there were definitely some AI messaging here and there like, oh, enabled IDPs and things like that. But it wasn't an AI conference just yet, not just it shouldn't be. And it shouldn't be. Although I did notice that if you if you went and talked to some of those vendors, right, and said,
00:14:33
Speaker
Hey, AI is here and is penetrating the thoughts of the KubeCon attendees and vendors. What are you doing with it? They would probably come back and tell you, they do have a project going on or they have it tied in here and there. At least what I noticed is that if you kind of poked at what people are doing about it,
00:14:54
Speaker
there's active stuff. So I think maybe we'll see, you know, next KubeCon. There are a lot more talks that have, you know, here's what we're doing with this model or that kind of thing. I think it was definitely top of the mind for all the vendors. And then I think next year you'll see
00:15:11
Speaker
not just talks, but also like productized things around AI that is actually using, say, an LLM or building their own LLMs for Kubernetes. So I think that's going to happen. Like, just one of the vendors on the booth, Oh, Lama. I think that's a really cool name, Oh, Lama. And they had like awesome Lama stickers that they were giving away. It was just like a three or four year old company. I stopped by their booth. They were surprised that CNCF used them as an example to talk about AI on Kubernetes. But they're like,
00:15:41
Speaker
they allow you to download all the LLM models on your laptop and then you can just run prompt engineering on your laptop instead of sending all of your data or all of your prompts to OpenAI or other vendors that post these models for you. So they are really hard on the experimentation phase right now just trying to see how this will work but right now they have a solution that is pretty cool. You can install it on a Mac or a Linux box or on a Kubernetes cluster and get local access to LLMs without having to send things over.
00:16:09
Speaker
Yeah, you know, it's it's funny you mentioned llama and because I you know, I was part of the data and Kubernetes Yeah, day one event and that was a really great event. They did a really good job You know putting those on and kind of making it feel like a separate thing that you're at but there I was at a talk about vector databases and And it was it was really well done but I sort of learned more about the community and things like hugging face dots and
00:16:38
Speaker
And I don't know what it is about the names that people are coming up with in the AI community, but I kind of love it. Some of them are sort of terrible and funny, but then like hugging face and llama and you know, I don't know. There's just something about people are having fun with it. And I don't know if it's just like, because we're so early on, hasn't like marketing hasn't grasped their hands on it and named it something like super boring. Traditional corporate marketing, yeah.
00:17:05
Speaker
Um, but I kind of like where that, you know, whole community is right now. Like just hugging face is the community. Like, I don't know. If you, if you didn't, if you're not familiar with hugging face and that's the first time you're hearing it, you're probably like having the same reaction I did, which was like, that that's the name of it.
00:17:22
Speaker
Okay. I don't know. So maybe it's just me, but no, no, it was like, it doesn't clearly say what it does, right? So like, even when I heard about it a few months back, it was like, okay, let me Google what this is. Like, why is everybody talking about it? It's like a GitHub for your models. Yeah. Just in that one line description, but check it out. Like it's free to use. You'll see all the models and you can download and start using them. So
00:17:45
Speaker
Uh, do it. Yeah. Yeah. And he might as well just be named like banana peels. That might be slippery. Um, there's one model on there that I actually saw on a thread on, uh, Twitter, which, um, is, is about, um, it's basically a model that you can run
00:18:06
Speaker
And it looks, it uses your camera and in real time has David Attenborough narrate you. So like based on what you're doing, I'll post the link to the Twitter in our show notes because the video that this tweet has is fantastic, but it uses your real time. So if you like,
00:18:30
Speaker
you know, make an expression with your eyebrows or take a drink of water. It'll do it in sort of like a planet earth style way. And it's hilarious. And these are the best things that AI can be used for. Really? Yeah. That's just, that's just me. No. And I think talking about like continuing our team's discussions, right? Yeah. For the show, bring us back on track, Bob. Trying to do, trying to.
00:18:58
Speaker
A lot of platform engineering, like IDPs, like one of the vendors that I found, I forgot their name, but they took a different approach on IDP. Usually whenever we talk about internal developer platforms, we talk about creating golden parts for developers to deploy resources, right? Just do things this way. But they had a different view. Instead of that, they were focusing on the you build it, you run it philosophy. They're like, oh, you built X number of services. Now you are on the hook for,
00:19:26
Speaker
managing them on a daily basis. So they gave like a cool dashboard, which aggregates like PagerDuty alerts, Grafana or Prometheus metrics, it just aggregates everything gives you a neat dashboard, like how all of your services are doing what needs to be fixed. If you had any specific log events, they'll show it to you. So like every morning, first thing in the morning, the developer can just log in, look at all of their running services, figure out what needs to be fixed, and then they can like go and
00:19:52
Speaker
continue doing or pick up another Jira ticket to work on. But that was a cool interpretation of IDP for sure. Yeah, this is like platform engineering, going to a level where it's like, you want to build it, you want to run it? We're going to give you some training wheels, right? We're going to try to make it as easy and accessible as possible so you can succeed at this. I don't have to run it for you. Yeah. But there are a lot of those traditional IDP vendors, right?
00:20:20
Speaker
as an admin, you can create blueprints on how to create like an S3 bucket or any object storage bucket, and the developer can choose from a dropdown like, oh, I want to create a blob storage bucket in Azure instead of S3 and AWS. And the administrator can configure or use APIs from the cloud providers or Terraform and configure everything on the back end. So for the developer, it's really easy. So I think platform engineering, I know it has been a buzz for a while, but it's getting real. There are actual tools that people can start using.
00:20:49
Speaker
Yeah, I feel like my take on platform engineering was yeah, it was still a big topic. And it was still part of the conversation for most people. But I think we're still in the process of figuring out what's the right way to get there. Like, not that it's not real, it is real. And people are kind of figuring out now, okay, well, how do I how do I concretely get there? And that's where I think the the blueprints come in. That's where I think, like, how do you actually build the platform? And how do I do it?
00:21:15
Speaker
in XYZ manner. I think another part of this, which we both put in our notes here, which was the platform engineering maturity model, which was from the application delivery tag. Yeah, that was very cool. We'll put a link in the show notes, but basically it is a white paper from the application delivery tag. I think I'm
00:21:43
Speaker
Yeah, that's true. Like app delivery, technical advisory group, they did a white paper and now they followed it up with like a model where they have different stages that your organization can be. So kind of like the Dora metric reports that come out, like, okay, your organization can either be in the provisional operational scale level or optimizing state of platform engineering. And then they look at different aspects from like, how do you get investment for platform engineering? What's the adoption? Like what are the different kinds of interfaces people can use our operations being handled and how do you measure things?
00:22:13
Speaker
And they rate across this complex matrix that we'll put a link in the show notes, but they tell you where you are, if you are doing these things when you're talking about platform engineering. So again, I think it's more about giving people yardsticks to measure themselves again. It's like, okay, I thought I was doing platform engineering the best I could, but
00:22:33
Speaker
I can be doing things better. So I think that that's what this maturity model will help bring to the community. They have a couple images in that white paper too, that I think are really helpful. One is the capabilities of platforms, find that in the page contents, which is really the relationships between what they talk about as products, platforms and capability providers. So it really kind of, I think puts into a nice concrete example of like, when someone says platform,
00:23:00
Speaker
What does it actually mean? And the northbound and southbound are actually fluid interfaces in the sense that a platform's not the same thing and it won't be the same thing no matter probably where, which organization you look at it. It'll have some common pieces, identity, data, artifacts, those kind of things, observability tools. But then there's north and southbound stuff. And then the second one was sort of a graph based on
00:23:27
Speaker
where you are in that maturity model. And one thing I took out of this paper was that in every case, it kind of says that platform engineering is a slow start.
00:23:42
Speaker
Right. And I, and I was like, this is, this is a very valid thing, but, um, I also want to say that a lot of things are slow starts, right? Even if you were like to say how you're adopting DevOps, right. A lot of times it's some tools, some groups kind of advocating, some champions kind of pushing, and then you kind of cross this, this, uh, this boundary. So, um, find both of those graphs. Uh, I think they're really useful.
00:24:08
Speaker
I haven't looked at that graph, but I'm assuming that, okay, once we go through the slow thing, it just becomes a hockey stick graph and like, let's go. Come on. Not so much hockey stick exponential growth, but, um, yeah, it does take off after a certain kind of crossing there. Anyway.
00:24:31
Speaker
No, I think one of the other things that I like while talking about platform engineering, right, talking to the actual practitioners or operators, as you mentioned, one thing that they're struggling with right now is that multi cluster, multi cloud thing, like even the
00:24:45
Speaker
There are like the 100 level examples like creating object storage bucket. That's good. But then what about the hard things like how do you manage your access credentials across all of these different clouds and across all of these different clusters? How do you do that? I think that's one of the challenges that a few people that I spoke to are facing and they're like, let's solve those. And then we can go all in to platform engineering because
00:25:08
Speaker
Many of these organizations have some form of IDP built over the years which used to help developers provision VMs and now they have evolved it to provide Kubernetes clusters or namespaces on demand and they want to move towards more of that cloud native platform engineering stack but then I think once we solve some of these
00:25:28
Speaker
day two challenges. I think that's when it becomes like super real. Yeah. And that boils down to like, what's the real value you're providing, right? And, and, and the graph, I think that I was talking about also kind of talks about value, right? Once, once you're providing enough value and that number is not a thing, right? It's, it's different everywhere, but, uh, your users will basically be, be pulling instead of, um, resisting, right? And that's that flip that they talk about between them.
00:25:55
Speaker
I think it's the operational and scalable layers. Gotcha. And I like that Spotify likes to share these success stories, right? I think the one blog that they published like two or three months back, they were like, 96% of their developers are already using the backstage based IDP that they built. And they didn't even market the IDP to all of these different developers or internal business units. People, once the snowball effect started, they had to get
00:26:21
Speaker
they had to put in some work to get the snowball rolling, but then once they did, I think it created an avalanche and like, yeah, a lot more people signed up and that's the default way inside Spotify right now. And I'm so happy with myself that I got the whole avalanche thing going. All right, let's switch gears a little bit. Actually, the couple of things we have listed here was application platforms, which kind of just ties into the platforms things that we're talking about. Security, always sort of a big topic.
00:26:50
Speaker
when it comes to KubeCons and where we are just in the world with technology. And then multiple clouds, multiple clusters. I know it's weird, I pronounced both words, but multi-cluster, multi-cloud. I think we're also sort of a parallel topic to a lot of different talks. I saw a lot of people describing their architectures using multiple clusters and how they're doing that with different tools and different applications and kind of
00:27:20
Speaker
Uh, those kinds of things. So, uh, I don't know about you, but those were the other couple that I had. No, I think, yeah, agreed. Multicluster for sure. And security is always there, right? Like I can't ignore security. Like if you do that, that will be not, not great. Uh, I know in the news section, not great was not the way a lot of people describe it. It becomes a resume generating event. Like if you mess up on security, like nope. Or as it played resume depleting.
00:27:48
Speaker
Ooh, nice. Depending on how that goes for you, personally. Okay, so let's dive into an article or two while we still have some time left. How about you go first, Bobbin? We'll be right back after this short break. As long time listeners of the Kubernetes Bites podcast know, I like to visit different national parks and go on day hikes. As part of these hikes, it's always necessary to hydrate during and after it's done.
00:28:18
Speaker
This is where our next sponsor comes in, LiquidIV. I've been using LiquidIV since last year on all of my national park trips because it's really easy to carry and I don't have to worry about buying and carrying Gatorade bottles with me. A single stick of LiquidIV in 16 ounces of water hydrates two times faster than water and has more electrolytes than ever.
00:28:42
Speaker
The best part is I can choose my own flavor. Personally, I like passion fruit, but they have 12 different options available. If you want to change the way you hydrate when you're outside, you can get 20% off when you go to liquidiv.com and use code KubernetesBytes at checkout. That's 20% off anything you order when you shop better hydration today using promo code KubernetesBytes at liquidiv.com. And we are back.
00:29:11
Speaker
Yeah, sure. I'll start with one of the big windows in the ecosystem, right? Like Microsoft. Obviously, Microsoft announced a lot of enhancements to AKS, not just during the week of KubeCon, but also Microsoft Ignite, which was last week in Seattle. But I just wanted to highlight two things that are relevant for our audience. Azure Backup for AKS is now generally available. They announced this as an alternative to just backing up Azure disks. So now this is a solution that's available in the Azure UI or the Azure portal. It is based on Valero.
00:29:41
Speaker
It's a simple extension that you enable for your AKS clusters, and then you can just select what namespaces you want to back up and it backs up not just your TVs, but also all the different Kubernetes resources that you might have running. So a good tool to use, simple to install if you are already using AKS or if that's your one cloud solution that you have for Kubernetes.
00:30:03
Speaker
And then the second thing from Microsoft that caught my eye was a new operator that they like to call CATO, which stands for Kubernetes AI Tool Chain Operator. And it's another managed add-on for AKS, but it simplifies the experience of running open source AI models on your AKS cluster. So it automatically provisions the necessary GPU nodes that are required. If they're available.
00:30:26
Speaker
for sure. Yes, yes, if they are available. And then it reduces the onboarding time. So it helps you focus on AI model and usage rather than infrastructure setup because they handle all the orchestration of AKS and the worker nodes with GPUs in them. The one cool thing that stood out was
00:30:44
Speaker
They manage these large model files as container images, and they do this using the operators. Like if you have a model, it can be stored in your Microsoft Container Registry. As a container image, it gets pulled down. You run it on your AKS cluster by automatically provisioning those worker nodes, and then you can run inference through that. A really cool open source operator that Microsoft is now talking about. I don't know if they have been talking about it before,
00:31:10
Speaker
Now with the AI spin on everything, this is something that definitely caught my eye. Yeah, here's a very specific example from the article that I want to point out. We were talking about how the names are kind of funny. Yeah. Right. So they're talking about inference servers being one of the things they deploy. And the two that they described in this article is such as hugging faces 7B. Right. That might as well just be goobly glob and eglibity glob.
00:31:37
Speaker
Right. And then you have Nvidia's, which is like Nvidia's Triton interface and inference server, which is like, okay, well, marketing got a hold of that one. I had to bring that full circle because it was just like the absurdity in one sentence. That's true. Okay. How about you go next, man? All right. Since we're on the topic of AI, um, I did want to bring up cast AI's, um,
00:32:02
Speaker
on announcements around, it was two different announcements. It was automated workload, ride sizing and precision pack. I'll describe a little bit about what those two things are. But Cast was one of the ones, one of the companies that we covered in our cost.
00:32:20
Speaker
Yes, so now they got an additional Series B. I forgot when exactly that happened. They announced it during KubeCon. So that was 35 million Series B from vintage investment partners and existing ones.
00:32:37
Speaker
from creamed and uncorrelated. Anyway, the point being is they were a cost optimization platform, which there's a number of these out there, right? Cast being one that really focuses hard on this AI aspect and bringing into the fray, right? These new capabilities are, you know, I think the first one is a little bit self describing in the sense that workload right sizing is something that we're constantly
00:33:01
Speaker
you know, pulling at straws, trying to get right. You know, that's the idea that when you provision a pod, it has CPU memory associated with it. And what's the right scale out and CPU memory for each one of those pods and containers. And how do you how do you always make sure that that's, you know, right sized in the sense that for how it's running, where it's running, those kind of things.
00:33:24
Speaker
So this is basically Cast's version of being, I think by default, Cast already makes recommendations like every 30 minutes or something. And so this kind of pulls it into the fact that it can automatically kind of ensure that those workloads are running most efficiently and sized in sort of a cost-aware, holistic,
00:33:48
Speaker
approach to resource utilization. And then the precision pack is something that's a little bit different in the sense that it's based on scheduling decisions. So if you're familiar with Kubernetes, Kubernetes makes some basic scheduling decisions based on where it's going to actually put that pod.
00:34:09
Speaker
And there's a number of different ways it can do this, bin packing versus sort of spread and those kinds of things. Well, this kind of is cast approach to taking a cost aware and sort of it is a bin packing algorithm, but it will it will do things in such a way that
00:34:29
Speaker
workloads have to move around less is the way that I mostly took this on. So overall improves efficiency, and we'll do this based on sort of their cost optimizations and things like that. So again, I think these types of announcements and companies, this is like the very beginning, right, in the sense that we're, we're starting to see
00:34:52
Speaker
tools and models used for real time decisions, making those inferences based on real data that's being produced at an infrastructure layer, scheduling layer. Just because the complexity that Kubernetes and running tons of applications on Kubernetes is becoming, I think these tools are going to be vital for operating and managing. And to make an aside,
00:35:21
Speaker
based on this, right? We mentioned complexity, right? Complexity was a topic in some of the keynotes as well in KubeCon. And we'll also put in an article in here from the new stack that talks about Tim Hawkins sort of complexity budget that he talked about at the keynotes. I didn't watch, I didn't go to the keynotes fully. I watched a few out in the kind of four-year area by the show floor.
00:35:50
Speaker
The point being is that we're making Kubernetes do all the things, right? Queue all the things meme, your favorite one. And Tim's basic point was that
00:36:07
Speaker
developers and maintainers should consider whether it's going to make things a lot more complex because one of the things in the article I think is like, no is temporary, yes is forever, right? If we keep adding things and adding things and at which point do we blow the budget, blow this proverbial complexity budget? And I would be in the camp of like, we're already pretty dang close.
00:36:34
Speaker
It's about that opportunity cost. If you say yes to this, what are you saying no to in the future? What if that thing is more important? I like that there should be a balance for sure in what things you're agreeing to and what things you're saying no to. I like the quote, no is temporary, yes is permanent. I like that one too. If I find out who and where
00:36:55
Speaker
Where it was actually said I will make sure it's accredited but it's somewhere in this article I also want to like I also want to tie the fact that I feel like OpenStack went through a very similar thing where you know it So many projects kept spurring off. Yeah, and it got very very complex doing a lot of things and I hope my hope is that we don't like
00:37:19
Speaker
do the same thing to Kubernetes. I think this is sort of what Tim is sort of leading towards. I know CNCF likes to focus on Kubernetes as the platform to run AI. Even at the supercomputing conference, there were already vendors that were starting to talk about Kubernetes.
00:37:40
Speaker
Kubernetes was not built for running AI workloads. It was... Can it? Yes. I think the difference is we shouldn't bake in first order objects or projects that do AI. It's fine to use Kubernetes as a base to run your AI stuff on or create another company or project, but let's not make it part of core. Yeah, not part of core. It should be operators, custom resources, just like
00:38:10
Speaker
Let that be your reconciliation loop for anything and everything, but not make it heavier. So agreed. Yeah. So anyway, this is really good article. If you were at the keynotes or if you weren't, you know, go watch them. I think there's a lot to be learned about this type of conversation in the community, especially if you're getting involved or you are involved as part of
00:38:33
Speaker
projects or a maintainer, that kind of thing. We took an aside from CAS. I'll throw it back to you, which is my second. No, all good. I think the one last thing on CAS, the total money raise is $73 million, and I'm sure cube cost might be somewhere in that range as well. There's a lot of money being invested in solving these challenges.
00:38:59
Speaker
Maybe you can figure it out on your own, but if not, there are vendors that are working on it. So go check them out. Like I'm sure they have like a free solution that can help you do some basic analysis on your clusters today.
00:39:09
Speaker
Next, I want to talk about our friends at Red Hat. They had a few different announcements. They participated in multiple day zero or day one events. Red Hat Device Edge is now GA. Red Hat had a mixed edge computing strategy. They had different answers there. Now they're like, okay, recalling anything that's Edge related as Red Hat Device Edge, it's now GA. It has two different variations or deployment options. It is built
00:39:36
Speaker
on the open source community project called microshift which is like a lightweight open shift deployment and then edge optimized os that's derived from rel or at enterprise linux and then when you're deploying device edge you can either select rel and portman for very small deployments with static applications and only when you want to bring in the the orchestration capabilities of
00:39:58
Speaker
OpenShift or microshift, you can select that second deployment option and that allows you to deploy containers and manage them locally. But then a few other things that they made GA hosted control planes for Red Hat OpenShift on bare metal and for Red Hat OpenShift virtualization. Hosted control planes is actually a really cool project. I think there's an open source variation as well.
00:40:19
Speaker
allows you to create these multi-tenant clusters or create like a master cluster and then all the child clusters or worker clusters where you don't have to spin up or dedicate VM-based nodes or bare-metal nodes to run control plane components for your actual workload clusters.
00:40:38
Speaker
The control plane components actually run a spot on the master cluster. So you gain a lot of infrastructure efficiency. The spin up times are way faster. I'm a big fan of hosted control planes. I wanted to share that with the community. And then if you are using Red Hat OpenShift service on AWS or Rosa, you can now run virtual machines because now they support metal instances in AWS. I'm still having a hard time justifying why somebody would
00:41:05
Speaker
spin up metal instances in the cloud because they are expensive and then run virtual machines on OpenShift. On-prem, on bare metal, I completely get that. But again, they do say that they have a few customers that are asking for it. So OpenShift virtualization on Rosa is now GA as well. I think, or I could see the use case around development and things like that, or testing, right? When you want sort of that, like,
00:41:31
Speaker
full environment, so to speak. And I think it makes sense, right? They don't force you to get metal instances. It can just be a temporary work group, node group that you deploy with a machine set configuration, get spun up, you do your testing or whatever you want to do, and then you can scale down the cluster. So I think there's still some value, but just the cost might be expensive.
00:41:52
Speaker
Yeah. So, uh, you know, going back to the red hat device edge thing, I know we covered cube edge in one of our previous episodes, which is a much larger architecture for how to run edge and how to talk back to the, your core cloud and those kinds of things. Do you think this is definitely the start of what we're seeing from red hat when it comes to maybe building like a full
00:42:18
Speaker
a full sort of edge cloud to edge kind of capability that we see in the kube edge world. No, I think agree. They do have their multi-cluster manager in ACM or advanced cluster management. I haven't tried this out personally, but I'm assuming that if you have microshift running at these edge locations, you can manage all of these sites from that ACM deployment that you might have.
00:42:43
Speaker
There are some capabilities for sure that will help you to manage everything centrally, set security policies centrally using your advanced cluster security tool or ACS. I think there's definitely value, but I think they still want to keep the footprint at the end edge to be as low as possible. Because again, as we discussed some of the challenges earlier, I did was in the edge computing episode, these are smaller devices, not server grade, not data center grade. So you want to make sure that whatever you're running is lightweight.
00:43:11
Speaker
Yeah, and I think that was called Edged as sort of their little runtime in Cubedge. But I'm more curious because I think Red Hat with its sort of track record with management operations being something they're very good at in terms of
00:43:29
Speaker
usability of their UIs and deployment, those kind of things. I could see them offering a very specific version of Red Hat Device Edge that also has all these other things tied into it, like syncing data back and forth and having device twins and all these things more tailored to here's an architecture, but for Red Hat.
00:43:55
Speaker
I'd put that on my predictions, maybe in 2024. That's awesome. And then I know around the edge computing space, since you are on the topic, another vendor in the ecosystem, Zedda, Z-E-D-E-D-A, are from my friends in India, Z-E-D-E-D-A. Zedda.
00:44:14
Speaker
Yeah, they announced like an edge community service which is like they're supposed to be like industry first fully managed community service for the distributed edge. So it's trying to solve all the same challenges like unreliable security, lack of skill, IT personnel, undependable network connectivity. But then they are partnering with Kubernetes ecosystem like vendors from SUSE to Canonical to OpenShift, like trying to support those distributions on their edge deployments as well.
00:44:43
Speaker
But yeah, if you are looking at building out hundreds of edge locations, check out this new vendor in the ecosystem as well. They refer to some like Gartner studies that say 80% of our customer software will be running at the physical edge by 2028. So they have some analyst numbers to back it up. But yeah, a new service, that's definitely worth checking out.
00:45:06
Speaker
Yeah, absolutely. Cool. So, um, another big topic was the Kubernetes API gateway 1.0 is live. Uh, there was a whole bunch of great articles and talks and stuff like that around that. Um, if you're not familiar with the gateway API, uh, it's really, uh, I like to describe it as sort of like in the grass dot next.
00:45:28
Speaker
so to speak, but also tailors to both getting traffic into the cluster and enter an Intric cluster. Basically, it's sort of the next rendition of how services are
00:45:47
Speaker
Are are handled it's it's definitely the has a lot more features a lot more kind of like usability so to speak for how things are doing it's been in progress for quite some time.
00:46:01
Speaker
But this is officially 1.0, and there's a whole bunch of great articles that I won't even try to start speaking to, but you can start using it and having it be stable. The other article I had around gateway API is solo.io is one we've talked about in the service managed category and stuff like that. They announced their fully conformant implementation of Kubernetes Gateway API within the Glue
00:46:31
Speaker
So lots of terms going on there. But I think this is something that we consistently will start to see from the networking and services vendors is kind of conforming to make sure they're using Gateway API as others.
00:46:48
Speaker
Dude, 1.0 is a big deal. I think a lot of vendors will start paying attention. Even though I work for a vendor, but I didn't know anything about it, but I need to check it out. What is it? How does it work? Thank you for bringing it to the episode, Ryan.
00:47:03
Speaker
Yeah, absolutely. Okay, next up, security, right? Okay, let's talk about a couple of security vendors. Aqua Security announced additions or new features to their Trivi open source image scanning tool. They announced the ability to generate K-bombs or Kubernetes bill of materials, which are basically a list of all the different Kubernetes components that you're running in your cluster, including control plane components, node components,
00:47:29
Speaker
add-ons, including their versions and images. Now they announced the capability to do vulnerability scanning of the components in K-bomb. So not a huge feature, but I think just making things better and better. And then another new vendor in the Kubernetes security ecosystem, StackLock. I know people who are familiar with Craig McClucky. He was one of the original Kubernetes guys. This is one of his new startups after Heptio. So StackLock is focused on
00:47:59
Speaker
improving the security postures, not just for developers, but also for open source communities and for organizations that are building using containers. I really like their theme on the website. They are all into marmots. They are animals that you find. Marmots, yeah. Pacific Northwest, at least. That's where I've seen them.
00:48:17
Speaker
But yeah, they announced like two open source projects, one called Mindr, which has its own Marmot logo. But it's an open source platform that helps development teams and open source communities proactively manage their security posture. So if you're working with multiple repositories, how do you do the repo configuration and secure those? How do you do dependency and license management?
00:48:41
Speaker
helps developers by enabling them to make better choices and enforcing controls. And then finally, helps them with artifact signing and verification using this open source six store project. And then trusty, which is which is something that actually I found better for some reason. So trusty helps you
00:48:59
Speaker
As developers, you like to use a lot of open source libraries and sometimes there are these malicious techniques through which attackers can get into your environments or your application code called starjacking or masquerading or typosquatting. What they have figured out using this open source project is
00:49:19
Speaker
If you are trying to figure out if this is a good open source report to use or not, you can actually like they will help you score like give they will give you a score based on their own data driven scoring method where they look at package provenance information where they look at the github stars activities and figure out whether.
00:49:36
Speaker
a package which was supposedly just called Markt, M-A-R-K-E-D, but a new malicious package showed up as Markt.js, which made sense to some of the JavaScript developers, but then wasn't an actual package. It was just a malicious package that, again, JFrog published back in 2022. So, Trustee will help you choose the right solution or the right package to use and not allow these malicious things in your application code as you're building your apps.
00:50:06
Speaker
So another cool window to look out for. Nice, nice. So while you were explaining all that really useful information, I was in the background finding the Marmot funny video where, if you've ever been on, you know, seen the video of the Marmot standing up and saying like, Alan, Alan, Alan, Alan, and you go, Steve.
00:50:29
Speaker
Anyway, I put the link in here because that is a marmot. In case anybody was wondering, it actually was. And if you don't and never have seen it, click on the link because it's just hilarious. At least I think it is. Again, I'm just taking down notes to make sure I check all of these things out. Plus, you have Gateway API number one, and then maybe the marmot video. You watch the marmot video first. Cool.
00:50:58
Speaker
Cool. So the other ones I wanted to cover here was that crossplaying, which we just had on the show was talked about as an incubating project. I think it happened either September, October, but it was brought up in one of the keynotes. It's really exciting stuff for them.
00:51:14
Speaker
OpenTelemetry also is 1.0, which is a big update, and there's a lot of different integration happening there. I know the Prometheus project is getting to that point, and there's some others that have already integrated it seamlessly to get all the logs and metrics and all those things.
00:51:39
Speaker
all together. So really cool stuff there. Speaking of Grafana, we both put this one here, so you might as well cover it. There's been some sort of built-in updates for Grafana Cloud around Kubernetes cost management. Again, this light cast and keep costs like we covered on the show before, this is sort of Grafana doing its own thing and just kind of, I think, speaks to
00:52:04
Speaker
I think where we're at with operating and managing Kubernetes clusters. This is clearly top of mind. And it's funny because I didn't put cost as a theme. But it's definitely up there. Yeah, for sure. And I think the thing that I wanted to talk about around Grafana was
00:52:22
Speaker
They have had their Grafana Cloud, which is a SaaS offering, where they host the Grafana instance for you. But now they have a solution where you can install their agent on your Kubernetes clusters and ship metrics to the SaaS portal. So making it easier instead of you having to deploy and manage these instances, even if you're running a Kubernetes cluster on-prem, install the agent using a simple hand chart and it will connect back into your account and show you metrics for your cluster as well. Cool. So I think
00:52:51
Speaker
We might have not covered every single link we had in here. We will still put them all in the show notes. But I did want to end with a few of our much more general thoughts, right? We covered that Kubernetes is maturing. It's not just growing and focusing, but really what we're seeing is a focus, a double click on stability, security, and things built on top, right? Things like platform side of things. And we covered that a bunch today.
00:53:16
Speaker
Platform engineering is definitely something that people are thinking about how to expose this to their internal customers and or external, I guess. The other thing I wanted to mention here, which on top of the complexity, which we talked about already, which is retaining talent is tough.
00:53:34
Speaker
I found there was a lot of people searching for roles at this KubeCon, and it may just be economically where we are and where the market is. But this is a little call to action to any of our listeners. We would love to aggregate any jobs you have in the community space and put them out to our community and to our listeners. So definitely go and send those to us. If we find any, we'll make sure and put them on our Slack.
00:54:05
Speaker
We'll create a channel just for this kind of thing or for jobs that we have that you want to post. So yeah, it was tough to see that. And maybe it is a number of different things. But yeah, retaining talent seems tough these days. I know. And this talent is not cheap, right? Like people wish with the Kubernetes skill set. I know we have been on a growth trajectory and more and more people are getting started with it. But yeah, it's still far and few. I don't know how the phrase goes, but they are not enough. Not enough, man.
00:54:35
Speaker
I like the audible there. And let's hire people because I think for a long time we were trying to hire the Kubernetes sort of unicorn, which is the person who knew how everything in the Kubernetes stack worked. I think we're getting beyond that point. Let's hire specialists who know a part of the Kubernetes thing really well. Maybe that's security, maybe it's application delivery, maybe it's CICD. Let's start doing that more. I think someone knowing everything is someone who probably
00:55:06
Speaker
is going to be less useful to you in the long run anyway, because it's just impossible to keep up. Yeah. And it's just like, they'll be so expensive. Like, again, this is not related to Kubernetes, but with all the things that are going on with OpenAI right now and how they are signing letters and they've ousted the CEO.
00:55:24
Speaker
I see tweets from Mark Penny of the CEO of Salesforce on Twitter. They're like, if you are somebody who wants to leave OpenAI, send me an email at salesforce.com. Just openly share. Yeah, openly just approaching people like, come on, join me. We'll match your pay. We'll give you all the options that you need. And expertise in relevant or popular fields is important. There's no shame.
00:55:49
Speaker
Yeah, no shame, dude. But that makes me feel good as somebody who's trying to think about that new ecosystem, right? Like, okay, this is the next place to be. Let's just skill up. So not just around AI, but at KubeCon, even though we are talking about practitioners and operators, I did get a few questions around, like, oh, what resources do you recommend for me to get started?
00:56:12
Speaker
have the new guy that joins my team get started with Kubernetes, hands-on labs, or just training courses. I recommended all the CKA certifications that CNCF has. Kubernetes is the hard way, and then for sure Kubernetes bites the actual podcast, right? You can learn a few things. I did have someone tell me that they used a few snippets from our podcast in an interview. Nice. I'm hoping the Madmit video makes it as the next thing that people talk about in interviews. It's got to be number one.
00:56:43
Speaker
That's funny. Well, as always, please go and give us a rating wherever you can on your podcast. Join our Slack. The easiest way to do that is KubernetesBites.com. It'll be top center on the page there. And just definitely encourage people to reach out, send a message, review us wherever you can. Hit subscribe.
00:57:04
Speaker
if you're that kind of person who listens on YouTube. I know. Thank you. Our YouTube subscribers have been going up, not as much as our audio listeners are, but it's going up there, dude. Come on. More people need to hit subscribe on our YouTube channel. Thank you. Thank you for that. As always, this brings us to the end of today's episode. I'm Ryan. I'm Bob. Thanks for listening to another episode of Kubernetes Bites.
00:57:32
Speaker
Thank you for listening to the Kubernetes Bites Podcast.