anynines helps businesses migrate workloads to Kubernetes or open source based Cloud Foundry with their modular, infrastructure-agnostic platform solutions. The company has recently onboarded some clients with interesting constraints to Kubernetes with the company having to navigate challenges with European OpenStack providers. OpenStack remains challenging with three main issues: automation, operating OpenStack itself, and the hardware and its integration into the network.
“Kubernetes also cannibalizes as part of that ex OpenStack territory and becomes an infrastructure abstraction. Well, you lose a bit of the automation power if those APIs aren’t standardized on how to control the life cycle of their Kubernetes, but at least you get that infrastructure abstraction and you can schedule your parts,” says Julian Fischer, CEO and Co-Founder of anynines, at KubeCon + CloudNativeCon Europe 2022.
Key highlights from this video interview are:
- Momentum continues with onboarding clients to both Cloud Foundry and Kubernetes. Fischer describes the challenges they have experienced with onboarding clients to Kubernetes with European OpenStack hosting providers.
- Fischer discusses the three main challenges they see with OpenStack and to what extent they remain unsolved. He discusses the role Kubernetes plays in infrastructure abstraction.
- With small to medium enterprises which may only have a few Kubernetes clusters, there are many components to help secure your cluster. However, with large enterprises with many clusters security can be complicated.
- Fischer explains the difficulties of lifecycle management, particularly when enterprises have a hundred Kubernetes clusters and they have all different operators and different extensions.
The summary of the show is written by Emily Nicholls.
Here is the automated and unedited transcript of the recording. Please note that the transcript has not been edited or reviewed.
Swapnil Bhartiya: Hi, this is Swapnil Bhartiya and welcome to our next edition here at KubeCon EU in Valencia. And today we have with us, once again, Julian Fischer, CEO and Co-founder at anynines. Julian, it’s great to have you in person.
Julian Fischer: Great to be here, and great to see you again, after that many years.
Swapnil Bhartiya: It’s been so long. First of all, you rode on your bike here. So first of all, talk about that experience. Also, if I’m not wrong, you recorded something for the data on communities committee as well, while you were riding your bike.
Julian Fischer: I was actually giving the talk here in Valencia, in their recording, in a hotel nearby. I mean, riding a motorcycle to conference has a bit of history. I did that back in 2011, going from Sao Punto to Spain as well, Barcelona, which would be Conferencia Rails, I think was the conference, was a wonderful experience. And I wanted to connect to that experience.
One of my teammates, a younger fellow, he also wanted to do a motorcycle trip and I said, “Well, let’s combine this.” And we took the motorcycles and went a little deviation through the Pyrenees, and then heading to Valencia. It was nice, three days, much fun.
Swapnil Bhartiya: Excellent. So I think you had more fun than a lot of folks who were actually here at the conference. You are here at the conference. What kind of vibe, what kind of energy you are seeing here? Because first of all, we are trying to get out of COVID. Committee adoption is growing. So the community is also larger, but a bit different also. So just share your experience.
Julian Fischer: Well, first of all, I’m very glad that the venue is so big and I never had the feeling that people are getting too close or anything. I have many good conversations. Walking into people again is just awesome. I’ve been missing that. So I think some people might still be a bit uncomfortable, but I think we are slowly getting there and it has just been a good experience to just see other people again.
Swapnil Bhartiya: Right. Now let’s talk about, I mean, we have talked about your company all the time, but let’s just talk about anynines itself. How have you folks evolved over time? Any specific problem that you are seeing that you’re still trying to solve? Because you’re like, “Hey, this is still an unsolved problem.”
Julian Fischer: Yeah. First of all, the Cloud Foundry business is going on. We’ve helping clients to migrate to open source based Cloud Foundry, including our data service automation. That’s continued at full throttle, works like charm. In the Kubernetes area, in the Kubernetes section, we’ve onboarded clients where there’s some interesting constraints. So in particular, one client, he has connections to customers in the banking industry and after recent events in Germany regarding Wirecard, for example, the Control Institute Baffin is pretty strict and observing those companies very tightly. So they basically asked us, can we find a hosting provider in Europe, preferably even German that will go with them the long way of going through all the additional security requirements they have with their banking customers. And with the anynines platform being modular and infrastructure-agnostic was a challenge for us because the major US providers have been ruled out by the customer’s customers very strictly. And it felt like connecting to our original story where anynines started with being a public past provider on European infrastructure.
And to just repeat a little bit of history, we had tremendous problems with OpenStack at that time, and we’ve been reevaluating European OpenStack providers. And we can see that these problems actually continue. So that leads to our conclusion that Kubernetes with their Kubernetes products in the hosting sector, or they distributed automation there, it’s not perfect, but you can get a managed or at least a Kubernetes cluster also on those smaller providers. So it means that the hypothesis has been strengthened that Kubernetes also cannibalize as part of that ex OpenStack territory and becomes an infrastructure abstraction. Well, you lose a bit of the automation power if those APIs aren’t standardized on how to control the life cycle of their Kubernetes, but at least you get that infrastructure abstraction, you can schedule your parts.
And in this particular case, it also becomes very important that data on Kubernetes is a managed, something that you as an organization managed. And as we have operators for Postgres, for example, they are a new upcoming for other data services as well, you have a self-contained Kubernetes deployment in this case, and all you need is a Kubernetes SAMware. So there, the anynines platform adopted to that particular constraint, and so far works well. And it’s a good showcase on how a pure Kubernetes workload, including data service operators and how this is an important thing.
Swapnil Bhartiya: Any specific challenge when you talk about regulations, which has, you mentioned, it has to be European and preferably German, what other challenges that are there when companies do face? Because OpenStack in the early days, T-Mobile, they had their own cloud. Actually OpenStack become quite popular in Europe because of the whole hyperscale market was US-based earlier, but how much problem they have solved, which problem that you are seeing that are still there, where it’s not, “Hey, you flipped the button, move to OpenStack, move to Kubernetes and all your problems are solved.”
Julian Fischer: Well, I think the problems we’ve seen are twofold. Maybe threefold. The first layer is your automation layer. So if you use a tool that talks to infrastructure, in our case, for example, Gardner, Kubernetes on demand, a module that we use, there are maybe some assumptions about the underlying OpenStack in there. And that may not be in conjunction with the particular OpenStack you’re talking to. The next level is OpenStack itself. So OpenStack itself is a problem for many companies, just to operate OpenStack, not doing anything wrong. So it’s just that some things in OpenStack are problematic. And the third layer would be whoever operates OpenStack needs to know OpenStack very well. The hardware must be organized in a very particular way and integrated with their networking. That must be done perfectly.
What we experience is problems like use an API call and say, give me a router, and you don’t get a router, or give me a load bouncer, but you don’t get a load burner. So basic things like that, they fail. Sometimes these OpenStack based services are marked as beta, so that they’re working on it. But it’s interesting. You can go to China and get an Amazon-like experience, at least for basic computer and so on, on Ali Cloud, for example, but where is that replacement near Europe? We are still looking for it. So Kubernetes is a good guest as an infrastructure abstraction. And I think that’s an interesting takeaway that, I’ve been publicly stating the hypothesis for many years and it seems to be holding to this day.
Swapnil Bhartiya: Anything else that you are seeing is specific to, I mean, we hear a lot about, of course you focus on data services, but we hear a lot about security these days. We talk about high availability also. What are you seeing in that space, especially from anynines’ perspective?
Julian Fischer: Well, first of all, it really depends on how you’re using Kubernetes. Let’s say small and medium enterprises, let’s say at the smaller range, they often have only a few Kubernetes clusters. And I think you have a tremendous amount of components you can use to get your cluster secure, to do networking, to do storage and service measures and all that. And I think that the ecosystem is very, very strong.
If you look at larger organizations where there are a lot of Kubernetes clusters, and you need to make sense of a lot of them, there’s a big demand for, for example, lifecycle managing them. But because if you think about a hundred Kubernetes clusters and they have all different operators and different extensions and so on, I need to make sense of that. On the data on Kubernetes community, I gave small talk, for example, about how hard it is to do lifecycle management for operators. So there’s Helm, there’s OLM, there’s maybe something like Caral from the Vienzo Tanzu platform, but none of these technologies allow you to lifecycle manage all these components in an effortless way. So if you are an organization and you run into the situation where you have dozens or trending into the hundreds of Kubernetes clusters, that’ll create a lot of operational friction. It’s also something that I’ve predicted for years. And I’m really interested to see that more technologies will conquer that space.
Swapnil Bhartiya: Julian, thank you so much for taking time out today. And of course talk to us, and the best thing, as you also said in the beginning, was to be able to see each other in person, because the experience that you get when you talk to somebody in person is totally different than doing a remote interview. So that is also there. And I hope that this will become a norm now where we’ll see each other often and do shows like this to talk more about this topic. So thank you.
Julian Fischer: Absolutely. And thank you for having me here. It’s always a pleasure.