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

[THEME MUSIC]

CRAIG BOX: NASA managed to slam a spacecraft into an asteroid this week. Take that, two critically panned films from 1998. Hopefully, next up, NASA can throw something at Mercury to stop it going retrograde quite so often.

Why do we only ever talk about Mercury? Plenty of other planets go retrograde, I just learned, after reading the Wikipedia article on what that actually means. Spoiler to the astrologically inclined, I do not think it means what you think it means. I had to shut Wikipedia before falling down one of those rabbit holes about whether there truly is a 10th planet.

Anyway, they say it's a time of change. And that much is definitely true. More on that in due course. If you want to be the first to know, you should follow me on Twitter, @craigbox. Until then, let's get to the news.

[THEME MUSIC]

CRAIG BOX: Not one, not two, but three big CNCF project updates this week. First and closest to my own heart, the TOC has voted to accept the Istio project. In the announcement, Google DevRel lead and all around nice guy Craig Box was quoted as saying, "This is a significant milestone for Istio and its community, and we are thrilled to reach the next step in the evolution of the project."

Istio uses SPIFFE, the Secure Production Identity Framework For Everyone, which provides a secure identity to every workload in a modern production environment. SPIFFE and its runtime SPIRE have graduated this week. You can learn about the two projects in episode 45, with Andrew Jessup.

The final CNCF update is that the Brigade project has moved to the Archive. Brigade is an event-driven scripting engine for Kubernetes originally built by the Deis team at Microsoft. With low usage and no ongoing corporate support, the maintainers first sought new contributors and then to archive the project. One of the maintainers intends to work on the project as a fork, and so it may, one day, resurface.

Sysdig's latest Cloud Native Threat Report reveals victims lose $50 for every $1 cryptojackers gain. Hacker group TeamTNT, still not the Doom level designers, stole $8,100 of cryptocurrency using $430,000 of cloud resources. How many graphics cards can you get for $8,100 these days?

As we discussed in episode 188 recently, the war in Ukraine has spilled over to the virtual world, with over 150,000 volunteers joining anti-Russian DDoS campaigns using container images from Docker Hub. The threat report complements Sysdig's Cloud Native Security and Usage Report, which we discussed in Episode 169. You can download it from Sysdig's website for the regular price.

Supply chain security vendor Chainguard has announced Wolfi, a Linux distro designed for the container and cloud-native era. Wolfi, with an I, is the name of the smallest type of octopus. It was built to support Chainguard's secure distroless images and could loosely be thought of as "Alpine without a kernel". It supports both glibc and musl, the latter of which you have probably never said out loud. M-U-S-L, Mussel, like a small octopus might eat. Chainguard also announced Chainguard Academy, an open source interactive educational platform, and the general availability of their commercial Chainguard Enforce platform.

Internet protection vendor Cloudflare introduced workerd, a JavaScript who wasn't runtime, based on the same code that powers Cloudflare Workers. It aims to straddle the line between “microservices” and “functions”, allowing hosting of JavaScript as a web server or acting as a programmable proxy. Thousands of workers can be hosted on one machine, something they say just wouldn't be possible using containers or even multiple processors.

Also launched this week is Palaemon, an open source Electron app for Kubernetes cluster monitoring and error analysis. Palaemon was designed to help identify the cause of OOMKilled pods within a cluster by showing usage metrics for all of the pods on a node at their time of death. Palaemon was a Greek sea god and is also a genus of shrimp.

Google Cloud has introduced Custom Organization Policy for GKE which allows you to define your own custom constraints to act as guardrails for your clusters. You could say that all new clusters must use Autopilot or that Binary Authorization must be turned on to clusters with a production tag. Google's Organization Policy can be used alongside Gatekeeper or Kyverno, with one enforcing constraints on the GKE API and the other on your apps and deployments.

Finally, when migrating to the cloud in Kubernetes, carpool company Blablacar prioritized reliability and velocity. After a while, it became time to reduce their carbon footprint and, quite possibly, the costs of generating it. Project lead Sebastian Doido writes about how they started at 25% utilization of their fleet, and through auto-scaling and use of spot and preemptive VMs ended up with over 50% utilization of a much smaller, cheaper fleet. And that's the news.

[MUSIC PLAYING]

CRAIG BOX: Betty Junod is a vice president for product marketing at VMware, leading the developer relations, product, and technical marketing teams for VMware Tanzu. Welcome to the show, Betty.

BETTY JUNOD: Hello, Craig. So nice to hang out with you again.

CRAIG BOX: Thank you. This is your second time at VMware. I don't want to guess at or imply your age, but it was a much younger company when you first started. How different is the company from then to now?

BETTY JUNOD: I was a much younger person when I first started as well.

CRAIG BOX: Funny how that works, isn't it? The passage of time.

BETTY JUNOD: [LAUGHS] I know. It's so weird, the one thing constant. Let's see. I worked at VMware 2005 to 2013, so it was right after the EMC acquisition.

I was pretty early in my career. It was about 1,000-ish people, so we would still have these Wednesday lunches. I mean, it felt like a startup vibe back then. What's different now? It's what? 35,000 people? So just a sheer scale of people, size of the business.

It's not just a little server virtualization company anymore.

CRAIG BOX: That would be a challenge for catering.

BETTY JUNOD: Yes. I remember the time that we opened up the new offices when they built the beautiful new campus. And we tried to have a one day lunch, and basically it proved unwieldy. It wasn't possible. We were too big at that point.

CRAIG BOX: Your career has always been a combination of marketing and tech. Which one came first?

BETTY JUNOD: I'm from the Bay Area, grew up here. So if you think about it from what kind of company town, we're not really a company town, but our industry is tech. So first job ever was during the dotcom — you know, I came at the end of it, towards the bust.

CRAIG BOX: You can't be that old then, or else the company business would have been oranges.

BETTY JUNOD: [LAUGHS] Yeah. No. We were trying to sell some stuff online. Not oranges, though. But it was during that whole, like, when online shopping was first trying to take off.

Fun fact — we were offered to be acquired by Amazon, and our founder said no. Because what is this silly little book company out of Seattle named after a river? But have always been in tech, enterprise software primarily right after that first dotcom thing. And really VMware was my third or fourth job out of school.

CRAIG BOX: Going through school, was it you wanted to go and study marketing and just because you were in the area tech was the thing you ended up going into?

BETTY JUNOD: Oh.

CRAIG BOX: Or were you a raging programmer at age five?

BETTY JUNOD: [CHUCKLES] You know, the joke is I keep slowly making my way closer to being more technical. Actually, I was a failed finance major. Leave it at that. So it was always somewhat of a business focus. I interned at a stock brokerage, and I said no thank you.

CRAIG BOX: Fair enough.

BETTY JUNOD: Yeah.

CRAIG BOX: You said before you were at VMware in 2005. That was just after EMC acquired them for the whopping sum of $635 million. It really was the era of everything going virtual. It was the time of the first cloud providers. What change did you see in the eight years that you were there?

BETTY JUNOD: You know, it was really interesting to see when I first got there it was like Workstation was huge. The idea that you could have more than one computer on a single computer and the whole ESX server side was just starting to gain some traction. So during that time, what I saw was it being like this really — people were like, this is so esoteric and weird. What do you mean? I don't understand — to literally virtualizing entire data centers and saying like, I took an entire floor and turned it into one rack.

It was very physical, that people could see. I even had one of my in-law's friends — he ran IT at a utility in California — come to me and say like, oh, you work at VMware? You saved my bacon. And I'm like, that's a real quaint way to say it. But what do you mean?

So he explained this whole thing of how they could stop using an entire floor. So that's the kind of stuff I saw, was it being kind of this nerdy thing that some folks would be excited about because they want to run multiple operating systems for testing — I write software, I want to test it — to becoming things that they were just using it to consolidate a lot of their preproduction environments. And then the big thing was when people virtualized their first production application.

I remember that being a big jump. And after that you started going into virtualizing everything around the compute. Then I've seen a lot of the same questions come up with cloud and containers.

CRAIG BOX: The first server virtualization memory that I have, I was working in IT consulting in New Zealand early in the 2000. We had virtualization software installed on a machine to run both a Microsoft small business server and a terminal server at the same time.

And I was confused as to, well, how, when you come and sit down at this machine because you have to physically walk up to them and put CDs in them and all that kind of thing to upgrade. How are you changing between these consoles? Struck me as being a strange thing, and it's surprising how quickly we went from that to trusting that the system management software would be able to handle everything and you just never had to worry about physically seeing or touching a machine again. That period was very, very short.

BETTY JUNOD: The company has been around for 20 plus years, but it did take, I want to say, close to about the first 10 for it to really gain massive traction. I was there when it went public, and I remember the headlines being like "the billion-dollar company you've never heard of," because always the stuff behind the scenes. [CHUCKLES]

CRAIG BOX: Did that buy your first house?

BETTY JUNOD: No. [LAUGHS] I was too low on the totem pole then. [LAUGHS]

CRAIG BOX: Well, then in 2014 you end up at another small startup, which may have some potential here. It's called Docker. What were they like at the time?

BETTY JUNOD: Oh, gosh. I showed up. They were a little bit under 100 people brought on to do their first enterprise product launch, but then over time really focused on both enterprise and open-source. At the core of it, I ran product marketing, product marketing, and partner marketing, and all the marketing surrounding it. So just got to really spend a lot of time with the technology, the ecosystem, and the community, and Docker was the first time I worked in open-source. It's a really special and unique time.

CRAIG BOX: How different were the audiences between the two companies? You're dealing with IT groups and people who have come from legacy infrastructure at VMware. Are you eventually getting to those same people at Docker, or are we still dealing with the open-source types and the people who just want everything for free?

BETTY JUNOD: The people who are buying container platforms at Docker, the stuff that once you get to running apps and production, they were still the same people that you sold VMware to.

CRAIG BOX: Were those people evolving? Were they saying, I used to run VMs and now I'm going to move to containers, or were you getting people who are only doing newer Greenfield applications?

BETTY JUNOD: I would say that at that point it was still super early. So you would have folks in the CIO org, right? The traditional infrastructure teams because it's infrastructure.

I was starting to see this formation of — they would have these new innovation teams because even cloud adoption at scale wasn't happening yet in the enterprise. And so they would have these new innovation teams that would play with this container stuff, play with this cloud stuff, try to figure out how are they going to operationalize it before they stood it up. So I was starting to see that parallel org.

CRAIG BOX: We talk about cloud-native as a concept, which isn't necessarily things running in cloud but running them as if you would run them if you were. In those early days at Docker, were people running in clouds, or were people still running traditional environments in the data center?

BETTY JUNOD: Both, I'd say. What's interesting is banks tend to be quite forward in some of the technology that they use. But for many of those folks, a lot of it just has to run on-prem, and some of it is a function of industry. But then there were other companies that we would talk to that they were combining the cloud-native — the patterns, and the practices, and the technologies — they would couple that with the adoption of actual cloud as infrastructure. We have too much choice now, Craig. It goes in any direction. That's the thing.

CRAIG BOX: Yeah. You mean the conversation or adopting technology?

BETTY JUNOD: Oh, this conversation and the technology. [LAUGHS]

CRAIG BOX: Well, there's a lot of different paths that one can go down. You've been in virtualization and containerization at this point. So the real next step, of course, is service mesh. And so it's not surprised that you ended up working for a service mesh company.

BETTY JUNOD: Mhm. Well, if there's anything that's consistent here, is I basically like abstractions. And I like abstractions that are platform-level.

After Docker, I met Idit from solo.io at one of the DockerCons, and then went and worked there for a while, and really got into if you then start to abstract more stuff away, so that all you're really left with is the business logic, and then you have the networking sidecar. And now we're seeing more of that with OPA and all this other stuff. The more you abstract out, everything becomes very purposeful, the process itself, but also it can get a little messy because now you have many, many more things. So it's really cool.

CRAIG BOX: Do you think that the end goal is for people who are deploying applications to only deal with those high-level abstractions, and then we'll have teams of people dealing with the lower ones on their behalf?

BETTY JUNOD: I think that's what we're trying to get to. So if we go back to then developers, really they should primarily be able to just focus on the code itself and then maybe some interactions that they want their business logic to have with the next service. But then really let folks who are more specialized in other areas have to deal with that. You can make some requests on the instructions you want to have, and then specialists can further fine tune that, versus just make sure that the cycles can go faster that way.

CRAIG BOX: By this time, all the traditional IT companies are seeing the rise of containerization. VMware got big into containers. They acquired Heptio. What lured you back there?

BETTY JUNOD: Because I was there before, still had a lot of friends there. Periodically, I would chat with folks, and they're like, hey, you should come back. But it wasn't until Tanzu and the Heptio acquisition and the formation of Tanzu and some other acquisitions that company was making that I was like, oh, you guys are starting to do some pretty interesting things. And then also there was a team that I really liked working with, and they're like, come back, come back, hang out with us. And I'm like, how can I resist?

CRAIG BOX: We'll give you lunch but only on Wednesday.

BETTY JUNOD: [CHUCKLES] Not anymore. You can work from home. That was the selling point. [LAUGHS]

CRAIG BOX: And it wouldn't be long before everyone could work from home.

BETTY JUNOD: Exactly. [LAUGHS]

CRAIG BOX: Your first job on your return was working in multicloud. Is that still a thing? Are people actually doing multicloud, or do they just really want to know that they could as an insurance policy?

BETTY JUNOD: So I think multicloud gets defined in a few different ways. The multicloud solutions team that I worked on when I first went back was really about what we were doing on the VMware cloud side. So it's like we have our infra stack, and you can get that anywhere. There is a use case for that. There's a use case of multicloud for, I'm a big company, and I have different apps, and the different apps live in different clouds. That is still multicloud.

I think the big misconception out there is that people try to define multicloud as just one specific pattern. But my belief — and the one I keep trying to convince everybody of — is that it just fundamentally means you can just use whatever you want. That's the whole purpose. Like, you don't need to be dogmatic and say I'm always only one cloud or one this.

There's lots of people that use Google for very specific things with AIML and processing. But then they may have some of their other stuff in Azure or some of their other services in AWS. And they're all kind of working together because now what you're doing is you're getting specialized services.

CRAIG BOX: To some degree, that's nothing different than saying I have a data center and I have some different types of hardware in it that do different things. It's just the scope and scale of which is increased a little over time.

BETTY JUNOD: 100%. It's like not every server in your data center is going to have a high-end GPU. But some will. It's that same concept. It's just that we know that the cloud providers, some of them have very specialized services.

CRAIG BOX: There was always the idea that there was going to be regular price changing based on who's got more capacity available and that customers would follow those prices around and move their things. Has that idea just gone away completely because of data gravity and contracts and so on, or do you think that there's ever going to be a world where people are going to go different clouds for the same workload at different times?

BETTY JUNOD: I think there could be a use case for different providers for the same workload. It could actually be based geographically what they're doing. There are some companies out there that do do active-active in two clouds.

CRAIG BOX: Yeah.

BETTY JUNOD: If you look at PagerDuty, they've been very vocal about it. The idea of this, you're basically day trading your application across hyperscalers. I think that's the pipe dream for vendors to figure out how best to help a customer get portability. But who has time for that?

CRAIG BOX: The same kind of people who have time for day trading. I don't know that they're really going to make a lot of money on it, but maybe it makes them happy.

BETTY JUNOD: [CHUCKLES] Yeah. I mean, actually, you know what? Day traders, that is their full-time job. Right? Fair.

CRAIG BOX: You very kindly volunteered to come on the show after I made a comment a couple of episodes back about not understanding the VMware Tanzu platform. Should we dig into it?

BETTY JUNOD: Absolutely.

CRAIG BOX: No pressure. You got this.

BETTY JUNOD: [LAUGHS] I might know a little bit about it. I try.

CRAIG BOX: Excellent. So let's go back to the launch of Tanzu. I think of Tanzu as being VMware's Kubernetes division. Is that right?

BETTY JUNOD: What we say is it's our modern apps. The Kubernetes layer and Kubernetes API is a defining architectural part of it. But inside out, we actually define ourselves as a VMware modern apps business.

CRAIG BOX: How do you define modern apps?

BETTY JUNOD: Things that are using the modern ecosystem of stuff, modern patterns. There are microservices, et cetera. So it was a lot of that spawned off of Kubernetes, yes, timing-wise. Yes, Kubernetes containers and cloud, that's the intersection.

CRAIG BOX: Are modern apps necessarily cloud-native?

BETTY JUNOD: From a pattern standpoint, yes.

CRAIG BOX: But you mentioned before about the big GPUs and so on. I can imagine that the category is broad enough that that could include running SaaS and not operating things yourself for certain parts of your application.

BETTY JUNOD: Yes.

CRAIG BOX: Tanzu was announced just after the acquisition of Pivotal. Just like you, Pivotal was one of those things that started at VMware, went off on its own for a while, and then came back.

BETTY JUNOD: Yes. Oh, my goodness. Yeah. So there are some long-timers where we used to call it the three-layer cake at VMware. And the Pivotal part of it was layer two.

That layer two part spun out and came back. It's the filling of the cake. [CHUCKLES]

CRAIG BOX: For people who aren't familiar, what was the thing that Pivotal did and where does it now fit into the Tanzu cake?

BETTY JUNOD: It was spun off as Pivotal, but it included things like Spring, the whole Spring framework. Rabbit. We had RabbitMQ. We have Greenplum, GemFire, so that had a bunch of data services.

CRAIG BOX: Is it fair to say it's all the things relating to applications, and programming, and data?

BETTY JUNOD: Totally. Yeah. All the things that you would build apps with, and then it had the lab-side services. And then the Pivotal cloud-fronted part was the paths in which you would build them and run those apps.

CRAIG BOX: So Pivotal as a company was sort of ambivalent about where you would run the services? And VMware as a company was ambivalent about what the things were that you'd run that just provided the operating system stuff. So it made sense to bring them back together.

BETTY JUNOD: Mhm. Yeah.

CRAIG BOX: You mentioned the three-layer cake there. Psychologists always tell you that people like things in threes. If you're building a telephone automation system, you should say, "choose from one of these three options".

I'm going to try and build a nice mental model of Tanzu today. The initial blog post announcing it said that it was about building apps, running Kubernetes, and managing a Kubernetes estate. Is that still right?

BETTY JUNOD: Yes. I would say, though, on the management side it's not just the Kubernetes estates, but it's also managing the applications.

CRAIG BOX: So let's do an easy one. Managing the Kubernetes estate part was through Tanzu Mission Control. And as I understand it, that still exists today mostly as it was when it was announced.

BETTY JUNOD: Tanzu Mission Control still exists, yes. It exists as it is today. It is now part of also this solution we call Tanzu for Kubernetes Operations.

So Mission Control really being around the cluster management capability, and Tanzu for Kubernetes Operations also includes things like observability, load balancing, service mesh, and, optionally, runtime if you want it.

CRAIG BOX: You mentioned Tanzu for Kubernetes Operations here. That's one of the two main categories that you have today. The other one is the Tanzu Application Platform. Are these two things family groupings? Are they products?

BETTY JUNOD: We have a lot of technology within the Tanzu family of things. We've got various open-source projects and such, as well. What we've done is taken all the things that we have, and we've really grouped it around the unique groups within organizations as we have seen and talking to customers that are working with these modern technologies to build and run apps.

So what we're seeing is there are teams that are around the developer experience and developer platform side, and there are other teams around the runtime and management when applications are deployed. So that's really where we are with Tanzu Application Platform. So that's about getting developers productive, having the paved path to production, security from build all the way to deploy. And then Tanzu for Kubernetes Operations is about, how do I get consistency and configuration, delivery of the Kubernetes estate, and the management of the apps and infrastructure environment.

CRAIG BOX: Is it as simple to say that the former is for the dev team and the latter is for the ops team?

BETTY JUNOD: From a value proposition for the user, yes. However — I'm sure you probably see this, too — is that there is this new more of this platform team that's forming in many companies that are making decisions about cloud and cloud-native technologies. We're seeing a lot of this reside there.

The platform team may still be the one that's setting up the developer platform, but it's all in service of enabling the application teams.

CRAIG BOX: So if we think of the traditional VMware products underneath all that, that finally brings us back to the three layers that we're looking for.

BETTY JUNOD: Yeah. And so on the traditional VMware side, the folks that have been with us on the journey for the last 20 plus years in the data center, we've really spent time to bring Kubernetes to them because it's a critical part of infrastructure going forward. Now, many of them are also investing in other parts of the Tanzu estate that we've talked about because it's for on-premise state up. But like I said, we're also seeing the formation of these platform teams that are being set up in parallel to the existing data center teams. And they're the ones that may say, well, what services from these clouds are we going to use? And then how will we normalize them so that they're consumed in some way for these different teams? And so we're really talking to both.

CRAIG BOX: Would you find a traditional customer uses both of these platforms, or would you find that some teams will take the Tanzu Application Platform and run it on top of their Kubernetes of choice and other teams will want to take your Kubernetes Operation tools and use their development tools of choice on top of it?

BETTY JUNOD: One of our design principles is that we are modular and flexible. So we are designed to work on any Kubernetes. So our principle is any Kubernetes, any cloud.

In that sense, if you look at the existing VMware portfolio, that is just one of the places that we also care about. So a great example would be on the Tanzu Application Platform side we actually invested a lot of time to integrate with EKS, ECR, and the AWS ecosystem of stuff to help folks who are trying to build to EKS, be able to kind of get up quickly with all the developer tools, and the pipeline, and all that.

CRAIG BOX: That is kind of you. They need all the help they can get.

BETTY JUNOD: [LAUGHS] Well, we've heard that they would like some more of the dev experience in front. So that's we're working.

CRAIG BOX: You also have Tanzu Kubernetes Grid. As I understand it, that is the runtime that's for people who don't have a Kubernetes service that they're using already?

BETTY JUNOD: Yes. So a lot of folks on-prem will want a Kubernetes distribution. So Tanzu Kubernetes Grid is our thing for that.

We do find some customers also run that in a cloud provider. So they may run that in Google Cloud. They may run that in AWS, and then they'll also run that in Edge.

What this just allows for the customer is flexibility of choice. And if you have a certain estate where you would need consistency across different environments, they choose TKG. Other times, we also do see the on-prem stuff is TKG, and the cloud stuff might be a managed service. I mean, the world is our oyster.

CRAIG BOX: That said, the company does have the legacy of the VMware virtualization product. Am I running TKG on top of VMware, or do I have an option because I know that Kubernetes provides largely the same set of things in terms of application management if I no longer need virtualization? Can I run that directly on the metal?

BETTY JUNOD: I might have to double check with you on bare metal part, but I believe you can because ultimately it's whatever the customer wants because the deployment target is up to them. I'll say one thing. In my experience, even going back to the Docker days, is that operational run books on things is a thing.

Changing those is no small feat because you've got so much process, and automation tooling, and what have you. So in some sense, for teams to be able to adopt new things into their stack, the more you can work into what they have, in some ways it's easier for them to kind of operationalize it, get familiar with it, and be able to provide that to their users.

CRAIG BOX: One of the things that was announced early on in the Tanzu story was a thing called Project Pacific, which was the idea of being able to do perhaps two things — run containers directly on top of vSphere without having to run them on a VM infrastructure like you would in a cloud provider but then also using the Kubernetes APIs to manage VMs and manage the traditional vSphere infrastructure. How is that evolved since then?

BETTY JUNOD: Yeah. Actually, we just announced this at our conference, what? Been a couple of weeks now? vSphere 8 with TKG 2.0 in there, and it is the continuation of Project Pacific.

So I like to think about it this way. If you are vSphere-native, we are bringing Kubernetes to you in your context so that it is from where you are already running workloads as VMs. You can now instantiate clusters and be able to run containers.

Even in the vSphere management tools, what we've done is you can now manage Kubernetes containers and VMs. So in the context where you already are, we're bringing those to you. Conversely, if you're coming from the Kubernetes point on out, we have a set of tools that are starting from whatever Kubernetes you have. But also if you're going to end up running a container, potentially your deployment target is a vSphere environment. You can use a Kubernetes API to instantiate into the vSphere environment.

So I think it's a pretty interesting perspective to do it that way. So we'll just bring the thing that doesn't exist in your current context to you.

CRAIG BOX: One of the other announcements from your recent conference was the launch of VMware Aria operations for applications. That was previously known as VMware Tanzu Observability. What do you do as a product organization when you have a need which applies to both cloud-native and traditional legacy applications like Observability? Which stack does it fit in, and does that have anything to do with why the product was renamed?

BETTY JUNOD: The renaming, there's a couple of things happening there. We have a pretty broad management portfolio. So we have the whole stack of things that manage vSphere environments, we have things that manage native public cloud environments, and we have things that manage applications.

And so as that product organization was looking at that, if you look at big companies, they're going to have a little bit of all of that in their estate. So one of the things was they were really looking at technically how do they unify that. So there were some announcements around the new Aria hub, and that's using a cloud graph. So it's actually using GraphQL to query across many different data sources. How can I get all this different stuff, and present it to you, and then be able to give you insights on that?

So if you're taking a look at that, the Tanzu Observability, it made sense to have that be part of that portfolio. So in that sense, for TO specifically, it's a renaming.

CRAIG BOX: Naming is famously hard. You have both Tanzu Application Service and Tanzu Application Platform. Why do you have both, and how do I keep them apart in my head?

BETTY JUNOD: Product naming is literally one of the hardest things. Branding comes second. There's a lot wrapped up in it. Tanzu Application Service, that is a rename of Pivotal Cloud Foundry. Tanzu Application Platform is the new thing that we announced in January.

CRAIG BOX: I actually remember there were a couple of different divergent paths for running applications with the Cloud Foundry model. I think there was still the old, what came from Pivotal Cloud Foundry, and then there was also a new approach using the same patterns on top of Kubernetes. Is there a single best practice today for how you should have people build applications? And does it learn from the legacy of Cloud Foundry, or is it an evolution of that product?

BETTY JUNOD: I think you've seen these. There were some tweets going around recently about no matter what we all try to do in the ecosystem space that we're in, after a little while, all we're trying to do is rebuild Heroku. And the thing is people love PCF, right? Pivotal Cloud Foundry.

It allowed you to have all these apps, run at scale with literally very few operators to have to support it. That was the beauty of PaaS. It was a super easy experience for your development teams, and then it also handled all the stuff automatically inside the platform. That was a beauty.

I think that Docker had a PaaS at some point before it became Docker. So the thing that we all realize, though, all of us PaaS builders is that that gets you so far with app teams. And there's a certain point where it may be the application itself, or as you start to scale across different teams, that they want to customize something. That's where it breaks the model from PaaS, because PaaS is about you just use the super-prescribed path that's here and that's what you get.

So I think that's where the learnings is. Like, people still want that super simple experience for the developer as well as the operational scale and simplicity you can get. But what they want is a little more modularity, because let's talk about you work at a mega enterprise that's a multinational that has many lines of businesses. Each one of those lines of businesses has a bunch of applications.

Just think about your insurance provider, where you don't get insurance from the state — [LAUGHS] health care and services from the state. So in America, your insurance provider has lots and lots of different services, and sometimes those are all different apps. Each one of those teams are going to want something different typically. And so that's where we run into some challenges. And also with the model of Kubernetes being very extensible and being able to swap out different components with a huge ecosystem. This is where I think you can take the learnings from PaaS. If you look back at the PaaS patterns, it's like there are outcomes that people wanted. But what they want is enough stuff baked in so you have a guided path, but they want the modularity to put in their own thing.

CRAIG BOX: So if we rearrange the letters here, we can have Tanzu Platform Application Application Service, and then you can have PaaS again.

BETTY JUNOD: Stop it. [LAUGHS] I'm going to have to send that up to the naming committee. I'm sorry, Craig. [CHUCKLES]

CRAIG BOX: VMware and specifically the ex-Heptio team released a bunch of open-source projects. Some of them are still very relevant and useful, like Velero for backup. Some of them are part of Tanzu, like Antrea.

Tell me about the state of the open-source ecosystem and the development that VMware is doing there.

BETTY JUNOD: There's a whole bunch of projects there, and what you'll see is they're actually part of the commercial offerings that we provide. So that's actually where you'll see it. So Carvel is actually part of TKG. So to be able to kind of deploy and update applications, it's been packaged in there. Things like Cartographer are part of the Tanzu Application Platform.

What else? There's a whole bunch. Basically, we've built these individual components, yes, and we continue to innovate on them. And then they're available for the community to do with and play with or work them into their own technologies. But we also do the thing of integrating them and provide them as part of a packaged solution with Application Platform and Kubernetes Operations.

CRAIG BOX: I guess that's an interesting question then. You say Carvel does application deployment and it's part of Kubernetes grid. Where do you draw the line for something that's sort of in the middle there, where it deals with deploying applications but not necessarily writing them? It could be part of the platform. It could be part of the grid product. How do you decide where to put something like that?

BETTY JUNOD: Well, I think if you look at someone getting TKG by themselves, they may still want to be able to do the managing the package to the cluster. So that's still useful.

So when you look at something like the App Platform and Kubernetes Operations, it's looking at an entire workflow area for the stack. Right? And the stack needed. And then the runtime itself you can look at is separate, but there are some things. A runtime without anything running on it, what is that?

CRAIG BOX: It's a work of art.

BETTY JUNOD: [LAUGHS] It is, but it's so much better when you have something running on it.

CRAIG BOX: You say that. As an ops team, I'm going to sit here and say no one is sullying my perfect cluster with any of those horrible applications that might go down at any moment. I love it. Bring it on.

BETTY JUNOD: [LAUGHS] And then they're just going to complain that there's a problem. It'll be, it's not my problem. It's your application's problem. [LAUGHS]

CRAIG BOX: Not in this case, it isn't, because there aren't any.

BETTY JUNOD: [LAUGHS] They're not running anything. [LAUGHS] Oh. If a tree falls in the woods, does anybody know?

CRAIG BOX: It depends.

BETTY JUNOD: Depends. Where are you standing? [LAUGHS]

CRAIG BOX: Senior engineer answer, it depends.

BETTY JUNOD: No. It is the answer of the century.

CRAIG BOX: You're responsible for not just marketing teams but also developer relation teams. How do those two functions differ, and what are the things that your DevRel team are doing especially around these open-source projects.

BETTY JUNOD: Gosh. Developer relations in marketing, is it in marketing or engineering? I think that is like one of the big contested debates out on the twittersphere amongst all of our friends in the industry. But I think it's really about who is your primary person or your customer that you're catering to and what are you doing for them.

From a broader product marketing team, it is more around, it is about taking a product to market to a set of customers. And oftentimes that includes buyers and all these other bigger constructs around who's your buying center, all the different personas, et cetera. Developer relations is really super targeted on the individual user, the person that needs to have hands on keyboard and touch, feel, and use your thing every day, and what is their experience for them. How do you make them successful with the technology you're building and also understand the ancillary technologies around it, complementary technologies around it?

So it's funny because I've seen DevRel take many different forms across many different companies. It can be 100% dedicated to the products you have, and I want you to help get people onboarded into my cloud or into my service. It can be more on just teaching more on thought leadership, so that people go to your DevRel person as their teacher for all things in a certain technology space. But I think what we do here is a mix.

So we do a lot of teaching on some of these newer concepts about microservices, about, there's Knative, and there's all these other things coming out in development teams right now, all the buzzwords. And we help break that down for the user because we want them to understand it.

And then we also do things like how-to, like how to use the stuff that we make, understand Cartographer, understand Carvel. And then what does that mean for, if you do something a certain way today, if you used this open-source technology that we happen to provide, how is that different?

So it differs, but I think it's that DevRel is really focused on a very specific type of person.

CRAIG BOX: You've worked for a megacorp. You've worked for startups. Which are you doing right now?

BETTY JUNOD: Oh, my gosh. It's like a funny inception thing. I'm doing both. I'm at a startup inside of a megacorp. Tanzu itself is only a few years old. We're very much startup mode. Some of our products, they've been out for about a year. We're rapidly delivering releases and stuff, but we are inside of a 35,000-person company.

CRAIG BOX: You have gone back to the scale where you could conceivably feed everyone on Wednesdays.

BETTY JUNOD: Yes. [LAUGHS] We're all distributed, though, so it'll be virtual feedings.

CRAIG BOX: Will you be coming together at KubeCon?

BETTY JUNOD: Yes, we will be there. And I'm looking for recommendations on the best Detroit-style pizza, because between that and the apple farms. So cider and pizza, everyone.

CRAIG BOX: All right. So you're going to have to explain to me what Detroit-style pizza is.

BETTY JUNOD: It is pizza that is baked in a pan, so it's rectangular.

CRAIG BOX: OK.

BETTY JUNOD: Like a sheet pan, like a jelly roll pan. I think it's got a slightly thicker crust, and they put a ton of cheese. So someone talked about this wall of cheese at the end, you know, because it'll crisp up, melting crisp along the edges of the pan.

CRAIG BOX: In my head, Detroit and Chicago are basically the same place. Is that fair, or have I just offended a whole portion of the US?

BETTY JUNOD: Craig, you're fired. We're no longer friends.

CRAIG BOX: It's all right.

BETTY JUNOD: If I just said, is New Zealand and Australia the same thing.

CRAIG BOX: Well, it depends. When we are in New Zealand. Let's say this is the thing. We're 95% culturally similar. So when we are in New Zealand and Australia, we will fight to the death over the last 5%. But if you take New Zealanders and Australians and put them in the UK, or the US, or whatever, we're like, yeah, we're close enough. We'll get by.

BETTY JUNOD: [LAUGHS] So you could say it's the Midwest. So, yes. As a region, it is similar, but they're both very different.

CRAIG BOX: I've had pizza in Chicago. I don't believe I've had pizza in Detroit.

BETTY JUNOD: Oh, OK. So the pizza, I think that is going to be the 5%. It's a very different pizza, sir.

CRAIG BOX: OK. Well, I'm going to look forward to seeing some pizza reviews come out of KubeCon, even if I may not be able to enjoy it myself. And I'd just like to say thank you very much for joining us today, Betty.

BETTY JUNOD: Thank you so much. It was great catching up.

CRAIG BOX: You can find Betty on Twitter, @BettyJunod, and you can find every last bit of VMware Tanzu at tanzu.vmware.com.

[MUSIC PLAYING]

CRAIG BOX: Thanks for listening. If you enjoyed the show, now's the time to follow me on Twitter, @craigbox, where you can send feedback. If you didn't enjoy the show, it was at least nice of you to not enjoy it all the way to the end.

The show is on Twitter, @KubernetesPod, and you can reach us by email at kubernetespodcast@google.com. Check out our website at kubernetespodcast.com, where you will find transcripts and show notes each week. We'll see you next time.

[MUSIC PLAYING]