Cloud Native Platform Race and Helm’s Next Chapter | Live from KubeCon NA | TSG Ep. 966

Hotel no-till Holiday Inn. We're a cube con. You're watching Textron Gang.

Hey everyone, it's Allen Hummel. It is. Uh, well then it's Thursday and we're here at CubeCon, uh, continuing our coverage from, from CubeCon.

Uh, this one, you know how we do Dextron Gang? We actually record it usually the day before. So we're going to give you the updates from Tuesday, Wednesday at CubeCon.

Mm-hmm. Uh, I'm joined here by the lady in the middle, our friend Hope Lynch, who was a both past, present and future Textron gang member. Yes.

And on the far right, the Dean Mike Vizard, who's been putting in, I don't know how many steps you must have put in the last couple days. Thousands. Thousands, thousands.

Mm-hmm. Not call Sagan billions and billions, but thousands. Yes.

And, uh, well, billions and billions doesn't really call Sagan anymore. It's OpenAI's pledging money. Mm mm But that's a whole nother segment.

Used To be a big number now. It used To be a big number, but anyway, we are, we're here to talk about a couple of things. Mike, why don't you kick us off with the first one?

So there's this kinda underlying trend here at the show where people are really talking about is efforts to unify the management of it. We've had cloud native on Kubernetes for a while, and then we've had more workloads still run on virtual machines and mm-hmm. These legacy applications.

And the latest example is Mortis is here with an update to their cordon control plane. It's an open source control plane based on the Kubernetes API that they've now extended to manage virtual machines plus Kubernetes clusters. Mm-hmm.

If you go down to the Red Hat booth, you'll see them talking about Cube Vert on top of Kubernetes to run legacy apps on top of the Kubernetes platform. Mm-hmm. In the same time, you know, Broadcom is out there banging the drum for, uh, its VMware Cloud Foundation where the Kubernetes is kind of integrated into the platform and mm-hmm.

All these people are talking about the same thing. Essentially, they're saying, look, we can't have all these separate IT teams managing stuff. Mm-hmm.

And this stuff starts to scale. We need to unify the management of it, and here's an approach to do that. Mm-hmm.

Hope, let me ask you this though. Mm-hmm. Is that achievable, desirable, or is this just some sort of theater we're playing here?

I feel there is an aspect of theater to it, honestly. Because when you were talking, I thought about, uh, cube Vert Kubernetes, right. Sounding like simplification.

Both of those are very complicated. Right. So who are you simplifying for?

If you are just simplifying because you are using fewer tools, yes, that's an option. But if you're using fewer tools, but the tools are more complex, have you really gained anything? And in this, you know, run to simplification, uh, you could reduce down to a number of vendors where you're really locked in.

You don't have negotiation room for pricing. Um, and also with the Broadcom VMware shift, I think honestly there's a little bit of a scare tactic in capitalizing on the VMware pricing changes by Broadcom. Not necessarily that Broadcom might be the best alternative for some enterprises.

Mm-hmm. Yeah, I agree. I, you know, I I I, I have some thoughts on this.

Mm-hmm. First of all, look, for as long as Kubernetes has been out there mm-hmm. It's primarily run on top of a hypervisor.

Mm-hmm. Whether that hypervisor was VMware or, or, uh, what's the Red Hat? One EE uh, uh, V the red hat one is KVM.

Mm. And Right. But there's another one then with three initials, whatever.

Mm-hmm. Kubernetes has always run on hypervisor. It was, that was the default.

Mm-hmm. I think an option was bare metal. Mm-hmm.

Mm-hmm. We've seen K three, K four, whatever that comes outta rancher that can run on optimize for Edge and so forth. Mm-hmm.

But, you know, running, running cloud native on top of your hypervisor, you know, I remember doing reports 10 years ago, eight years ago. Mm-hmm. Well, you know, uh, density of Kubernetes versus, uh, hypervisor mm-hmm.

Versus hyper vi Kubernetes on top of hypervisor. Mm-hmm. And what was more optimized.

Mm-hmm. So that's not new. What's new is, you know, the old, the mythic quest in technology for the single pane of glass that allows us to manage all of this from one screen.

Mm-hmm. Now look, certainly VMware Broadcom's licensing model change mm-hmm. Has given everyone, and it's a huge market mm-hmm.

An opportunity to say, wait a second, this is a good time. Mm-hmm. To our options mm-hmm.

Not just purely on pricing to move from one hypervisor to the other, but to move from the data center to the cloud to move from monolith to, to, uh, to multi, not multi-threaded micro Microservices based architecture to move from to cloud native mm-hmm. From not cloud native. Mm-hmm.

And so all options are on the table. Mm-hmm. And so you've got all these vendors from the big guys like Red Hat mm-hmm.

To the Morant of the world. Right. Who are looking to try to be scrappy with their mm-hmm.

You know, recycled software. Yeah. Um, you're, look, you know, you're looking, they're all trying to jump in here.

Mm-hmm. Including the hypervisors, I mean the hyperscalers mm-hmm. Mm-hmm.

Amazon, Google, Microsoft. Yep. They're all coming out with, Hey, you looking at move over VMware, check out our hypervisor based, our virtualization layer that will let you run mm-hmm.

Cloud native on top, and I'm sure Amazon will have some working mans mm-hmm. Version of this single pane of glass that's going to give you 80% of the functionality for mm-hmm. 20% of the price.

You know, you make me, in your conversation, you make me think about another aspect of this. Right. So there is something to be said for simplifying something.

Mm-hmm. But the problem, just maybe visibility. Right.

Do they just need a better way to consolidate the information that's coming from the tools to understand the tools they have, how they're being used, if they have enough visibility across, they might not be in favor of so much consolidation because they find things are working, but they just weren't getting the right information. Yeah. Yeah.

I was talking to, um, Sean Amir over at tis, and he was saying that more customers wanna move to bare metal because especially these AI workloads, and they don't want that extra layer of VM if they can get away with it. Mm-hmm. But he said the problem is, is nobody remembers how to provision physical hardware of these companies anymore.

Like, there's only people who know how to do VMware. So that's what they use, because nobody knows how to make the underlying hardware work to say the Rapid jack Is gone. You, you just took the genie out of the bottle.

You know, what happens when you assume? Hmm. I think a huge mistake that too many of the vendors on this floor are assuming is that we're going to use Kubernetes to manage my AI stack.

Mm-hmm. And I just, I published an article yesterday about cracks in the invincibility shield of Kubernetes. Mm-hmm.

And I, I think a potential, a potential more than a bump, a potential roadblock in the horizon mm-hmm. Is that the decision is made by the market that an AI stack needs to be managed by an AI orchestrator or an AI platform, or, you know, an AI tool. Mm-hmm.

Not Kubernetes. Mm-hmm. And then we're going to get into a thing of Kubernetes is not the default mm-hmm.

For AI workloads. Mm-hmm. It happens to be the orchestrator we happen to have at hand.

Mm-hmm. But it may not be the one that we wind up using, As I wrote in my article. Mm-hmm.

We're one fan will snap away. Oh, Everyone is gonna be gone Uhhuh. Well, at least 50%.

At least 50%. What do you think? Well, I think that, um, this is a theater that appeals to the CIO because then it allows them to start saying, we can reduce the total cost of it.

Mm-hmm. But I think when you get down into the rank and file, there's, you know, engineers that have Kubernetes skills are not gonna give that up for something else. And same thing with the vm, where the, the IT admin who's not an engineer kind of likes virtual machines 'cause they understand how those work and mm-hmm.

I think that we're just forgetting the human factor in all of this equation. And people have preferences and biases. And as you said that, I also thought when someone is thinking about their career trajectory, their higher ability, they want to continue to have hands-on experience with these tools.

So if the company starts to say, we are gonna sweep these away, and they know that this is something that they need to have, you know, they may make decisions to walk or they may push back hard enough that the organization decides to shift gears. I I, I think they start looking around at the one ads to see who else is out there that appreciates me more than these guys. Exactly.

Mm-hmm. Yeah. Mm-hmm.

Yeah. Yeah. Well, so Think, think about CIS admins before Cloud Native.

Yeah. Waterfall people pop. Mm-hmm.

Right? Mm-hmm. Well, you know, but these days, you know, you became a VMware administrator from assists admin 'cause you could get yourself a raise with a certification.

Mm-hmm. Well, now you can get yourself a raise by becoming a platform engineer and get that certification and get another raise to go with that. And I'm, you Know, well, I I do, you know, that's something we haven't spoken.

Mm-hmm. Well, you mentioned it earlier on Hope mm-hmm. Which is the influence of platform engineering on this whole thing.

Mm-hmm. But, you know, it's funny, when, when platform engineering first burst on the theme scene, it was very much a discussion about managing Kubernetes, managing that cloud native stack, if you will. Mm-hmm.

Mm-hmm. Today, platform engineering is all about IDP. Mm-hmm.

But the IDP is that pick list, that catalog almost like A-C-M-D-B used to be. Right. But it's that pick list and that catalog.

So if you say, I want to run containerized applications and I want to use Kubernetes, that platform engineering team will be the one who has prescribed those technologies, how they work, how they connect in the enterprise. But without that platform engineering team, it's a little, probably a little less, you know, elegant and smooth, but can still be done. It, it'll be interesting to see platform engineering, but I, I look, I'm just sitting here waiting for the other shoe to drop mm-hmm.

Into 'cause 'cause here's the thing. For the last two, three years after every Cube coupon, I always do my wrap up article. Mm-hmm.

And I write, is there anything on the horizon that could stop this juggernaut? Mm. And this is the first year I feel like Yes, there might be.

Oh. Mm-hmm. Well, Good.

Yeah. I think we're good. There's a war on out there.

We'll see how it turns out. Yeah. There's Always a war on there.

Always the state of humanity. Hey, we're, we're here in Atlanta at Cucu, we're gonna take a break, we'll be right back. Mm-hmm.

You've earned it. The spotlight, the responsibility, the weight of teams, companies, and entire industries fall on your shoulders. Lives depend on your decisions, your home life included that work you are protected physically and digitally.

Nothing gets through your team without a fight. But in a globally connected world, everyone sees you, including those who mean to cause you and your organization harm. And now home your sanctuary attackers see an opportunity, your digital front door is wide open.

And what compromises your home can breach your border. Because the Devil's greatest trick isn't targeting your workplace firewall. It's convincing you that your personal life isn't at risk.

Black Clerk, digital Executive protection, defending the new attack surface your personal life. Hey folks, we're back at CubeCon and we're gonna talk about Helm Helm four has been announced, and this is kind of like the Package manager. That's the first one that was out there, I think.

Anyway. I think there was maybe one earlier, but this is clearly the one that's most widely used and mm-hmm. After 10 years it's getting an upgrade.

The last upgrade was six years ago. And well, helm is kind of everywhere, although some PE people use operators and there's other ways of packaging software. But, um, you know, I was talking to Matt Butcher who actually wrote the original help, Huh?

From Fer Mine, isn't he? Yeah. He's over at Ferry Mine.

And he was saying, you know, I laughed and I said, 10 years ago, did you think that this would still be around and that this would be so widely easy? He goes, no way. I was just trying to solve an immediate problem we Had.

Well, I I think they thought operators were going to replace Helm. Mm-hmm. Right.

Because the whole concept of operators and, you know, you didn't need help, But, well, I mean, I think operators are cool, but they're only like for one piece of software and people use Helm that do multiple things in multiple packages. Mm-hmm. So I think that there's, people are using these things and mixing and matching them as almost as a personal preference in some ways.

Mm-hmm. Or what's ever convenient and what's ever handy, but mm-hmm. Mm-hmm.

I don't know. What are you seeing? I am, I am seeing that Helm Yes.

Has retained its popularity, but a lot of people have complained about age. Right. Because it still does have, uh, probably, I, I think it's the second biggest community in the CNCF behind Kubernetes, or maybe it's above Kubernetes.

It's not anymore. It's not. Hotel is Third.

I think OpenTelemetry Third is Argo, But it's still, but that's still pretty high. It's still at the top 10. That's still pretty high.

But, you know, the 10 years it's showing its age. CNCF is, you know, doing a new initiative Right. To basically revitalize Helm.

And I think that is one a testament to, there are a lot of people who don't want to necessarily shift Helm is the default for so many mm-hmm. Because of that CNCF is saying, well, you know, we have to put some strength behind this and do something, uh, to keep the wheels from falling off, so to speak. And one of the things, you know, if this could be fixed, I wouldn't mind, but in Helm you use YAML to modify YAML to do something, to do something else with yaml.

Right. That, that layering can be complex and confusing. Um, I'm, I'm, I would say I'm still a home proponent, but there are, there are definitely a lot of improvements that can come.

So, So this is the magic trick they're trying to pull off. They wanna rewrite the API mm-hmm. While trying to re preserve as much backwards compatibility as they can.

Mm-hmm. And then they wanna use, uh, web assembly or a K Wasm as the framework for the plugin. So I don't have to write the same plug.

Well, I I, And over again, that's Matt butcher's take on it. 'cause he, he has a horse in that race. He, he said he had nothing to do with that particular decision.

I asked him about that. Oh really? You did?

Okay. He said, he said that, he said that was, I'm not, he's not directly involved in this project anymore. He's just kind of hanging around the edges of it.

Mm. But At least the Epstein files. But, but be that as it may, um, I think that this is gonna be hard.

I think It is going to be very hard. And I think we're gonna wind up with two Helms. We're gonna have, you know, helm Classic and, and the new helm.

I don't think that's necessarily a bad thing though. Right. Right.

Uh, that's almost taking a mode, mode one, mode two approach. Right. If someone wants to or needs to stick with the prior version, fantastic.

But if you release yourself from, uh, the shackles of all of the backward compatibility and maybe for new installs, people see that there's, you know, there is a better way. That's fantastic. It almost reminds me of when Microsoft was like, oh, you know what?

We need to create a new operating system. We are gonna start fresh because it is just too much work. It's not worth it to modify the old.

Right. We, we, you know, we've got to change. I think it will be a smart decision actually.

So, so I'm gonna pull the bandaid off of this scab off this wound. Mm-hmm. Here's the real issue.

Mm-hmm. You heard it here from Shimmy. First.

Shimmy said it, The real issue is the politics of what goes on with graduated projects in the CNCF. There's a reason why Helm hasn't been updated in all this time. Right.

You got one foundation with 200 plus projects to mm-hmm. To sit over as a mother hen. Mm-hmm.

And you've got some monsters in there, as we've said. Mm-hmm. Oel, Cobe, Prometheus, helm mm-hmm.

Argo and 190 some odd others. Mm-hmm. And they, they can't, you can't possibly, I don't care how much money they're bringing in mm-hmm.

You can't possibly do Right. By all 200 of these. And you pick your bets and you pick your places and they leave it to some of the projects to kind of self govern themselves.

Self-manage themselves. Especially ones that are graduated. Helm's a graduated project.

And, and so you got a guy like Matt who is no longer part of that project. 'cause he's busy with Watson. Matt, the other Matt, Matt Farina over at SUSE is the primary maintainers.

Yes. Mm-hmm. Yeah.

So you've got, you know, you've got people coming and going who helped, uh, really put this on the map. Who would, who are the driving force? 'cause make no mistakes.

Open source isn't the, the, you know, the shiny hill on the city with all those points of light. Mm-hmm. There's generally a handful of people who drive a project.

Mm-hmm. Mm-hmm. And when you lose some of those people, you lose some of that momentum.

You lose some of that vision, you lose some of that. And I think that's what happened here with help, right? Mm-hmm.

Mm-hmm. I think to your point, I think a lot of vendors like operators better 'cause it gives them more control and Absolutely. Mm-hmm.

Mm-hmm. And, but, but no one wants to tell helm that because it's still one of the top tens and it's all of this stuff. And, and, and by the way, it's not just, this isn't a helm specific problem.

Mm-hmm. This is a, a problem with, I I've spoken to several other insiders on some graduated projects here. Mm-hmm.

Or some projects that actually have met the metrics to graduate mm-hmm. And have not been allowed to graduate because of politics. Mm.

You heard it here first. I'm not naming names, but I do have a deep throat. Outstanding.

Outstanding. All right. The rebellion starts tomorrow.

It does. It does. Now we'll fade out to the twin sons of tattooing.

Um, we're watching The Revolution will not be televised. No. We're watching, we're you're watching Textron Gangs.

Stay tuned. We'll be right back. Discover Textron Group, the epicenter of tech innovation.

We are your go-to for reaching it leaders and practitioners worldwide. Our secret impactful content that sparks awareness, engagement, and top quality leads with us. You'll access editorial websites, streaming videos, virtual events, custom content analyst research, and more.

Join our satisfied clients. Let's revolutionize your tech journey. Contact us today and tell your story to the world in the most powerful way with Textron Group.

Hey folks, we're back in. As Alan alluded to at the top of the show, we're gonna talk about OpenTelemetry, AKA hotel according to the cool kids. Anyway, um, there's a small company here called Buying Plane that has a new tool for deploying, uh, hotel collectors at scale, which I think is becoming a more pressing issue these days.

'cause we're trying to instrument everything, right? If I want AI to work, I gotta have telemetry data from everywhere. But now I'm getting overwhelmed by telemetry data and I can, and then if I put out a hotel collector, I need to maintain and update it.

And this becomes a DevOps project in its own right. Mm. So, so if you kinda look at this hope, um mm-hmm.

You know, what are you seeing with hotel these days? Is it gonna become the default? And, you know, it seems to be in the cloud everywhere, but on premise not so much.

And certainly not networking and security yet, but it looks like we're headed that way. I think, I think we are headed that way. But the, the road might be a lot longer right.

Than people are thinking. The, the basis of it, the principle of it, the promise of it is fantastic, but you have to get organizations to implement it the right way to standardize all of those things. And what I have seen is there might be solid adoption in certain pockets of an organization.

They are running, they are getting the data and they have the information. But then when you were saying, uh, OpenTelemetry for all, uh, a lot are not adopting it or are inconsistent. So that promise of having an enterprise or having an organization that is able to bring all of that to data together and get a comprehensive view probably is, is not as true as people would hope.

You are familiar with Cowboy wisdom. The phrase, the phrase, the phrase of cowboy wisdom is never mistaken clear view for a short Distance. Exactly.

Exactly. Exactly. Yes.

But yeah. But OpenTelemetry is fantastic. I I'm a proponent, but yeah.

So look, I give them my award for best, uh, handouts here. Not everyone could get 'em. Mm-hmm.

But the OpenTelemetry baseball jerseys that some of the guys have Oh, are hands down the best thing I've seen here in vendor wear. Nice. Um, but that being said, O hotel is huge.

It is the under the hood choice fueling the entire observability market. Mm-hmm. Mm-hmm.

Which, when you think about it, it's pretty amazing. Mm-hmm. And what's really nice about it is whether you're this little company who's, you know, making a little widget on it.

Mm-hmm. Or you are, you know, the giants of the observability space mm-hmm. Splunk, new Relic mm-hmm.

Or, or what have you. Mm-hmm. You all have a, a vested interest in keeping tel healthy.

Mm-hmm. Mm-hmm. Mm-hmm.

And I think that's when this foundation model works best. Mm-hmm. Right.

The cooperation Mm-hmm. Aspect mm-hmm. Makes the best.

There are still IT organizations out there that prefer the older proprietary collector they got from a vendor. 'cause it's easier to manage and they spend more time polishing it. Mm-hmm.

And then they, and then, so the vendors always look at 'em and say, we highly recommend that you deploy hotel, but you know, if you really want this other thing, we still offer it. Well, For an added fee. Yeah.

But, But I think the vendor under the hood has hotel in it anyway. Mm-hmm. Because why should they continue a parallel development check when hotel's there for them to put into, just Put a wrapper on it.

Right. Yeah. There's just, there's just pride and ego and extra money.

Mm-hmm. Mm-hmm. Mm-hmm.

Mm-hmm. Mm-hmm. Yeah.

Maybe, yeah. Monetize. But I, I think most of the observability vendors, whether they admit it or not, have made that choice to go to o hotel.

Yeah. Well, I wanna mention it's not just O Hotel. Mm-hmm.

It's Prometheus. It's graf. There's a lot of great observability.

Mm-hmm. Yeager, there's a lot of great observability, open source project. I Mean, from their, from their point of view, the more telemetry data they can collect, the better off they're gonna be.

So if I'm Until you gotta pay for storing all that. Well, That's true too. But if I gotta pay for the collector, I'm not gonna deploy it as widely.

So if I got a free collector, I'll theoretically instrument more, which will generate more telemetry data that the observability platform is gonna charge me more for. Yeah. So it all kind of is virtuous cycle in their minds.

Yeah. It continues to snowball. But I have seen in practice though, even with, um, OpenTelemetry and like you were saying, collecting more and more data, often a lot of it just sits there.

Yeah. No one actually uses it, does anything with it. It's just been an exercise of, oh, if I can collect more, I will have more.

And that's great. And people spend hours and weeks and maybe months Right. Making these configurations, but then they are only looking at just a few things.

Right. I think there is a also a need for training and maybe, you know, some, some type of framework that makes the decision making a Lot easier as well. So Cisco's trying to build an LLM for telemetry data, which would be kind of cool.

Mm-hmm. So we'll see how that goes. But I think the other part of the issue that comes up to your storage costing is so people are sitting there going, this stuff's expensive to store.

So they start tossing stuff, or maybe they try to move it off to S3 if they're ambitious, but you know, then they're sitting there going, well, by the time I get it back from S3, boy, that'll take a while. And so then I'm just tossing stuff and I'm hoping that whatever crisis emerges, that I actually have some telemetry data that's relevant to that. Right.

But you know, So I call that the Splunk conundrum. Right. This Has been the problem with Splunk for templates years.

Mm-hmm. It's greater collecting everything. Mm-hmm.

But where the heck, how can I afford to keep everything? Yeah. Yeah.

And, and so there's been a whole cottage industry of, of companies that come out and say, well, let us help you pick what you should keep, let us help stratify what you keep, where you keep it. Mm-hmm. How much you pay for it.

Mm-hmm. I remember there was a vendor, uh, Devo, right. That their whole, their actual GoTo market was, we'll let you store this stuff cheaper than Splunk does.

Mm. Mm-hmm. Yeah.

Well, I think we're gonna continue to drown in telemetry data. I think so too. Totally, Adam.

And I'm not sure if it's a good thing or a bad thing. Two letters for you. A i I, I think AI could, could help us here.

Seriously. All fitting aside AI could help us here in predictably guiding us mm-hmm. As to what we should store, when, where, and how.

Mm-hmm. Mm-hmm. All right.

I think we can end on that. Yes. Yeah.

All right. Hey, that's Textron gang for today. We've got a full day still of, well a half a day, uh, uh, Kubernetes coverage here on this beautiful Thursday.

And then Bonnie, keep the light on. I'm coming home tonight, hopefully, even if I have to drive. But, uh, for on behalf of Hope and Mike and myself, I hope you've enjoyed this text on gang.

Take care. We're out.