Sean McManus, TelecomTV (00:00):
Hi, I'm Sean McManus. I'm at KubeCon, and I'm joined now by Gabriele Bartolini and Simon Metson. And we're going to be talking about Cloud NativePG. Gentlemen, thanks for joining me today. Gabriele, let me ask you, first of all, why was CloudNativePG created, and what's its status today?

Gabriele Bartolini, EDB (00:18):
I think the best way to answer to your question is with two more questions. So why Kubernetes and why postgres? So we'll start with why Kubernetes. Kubernetes was created 10 years ago, and it's become the standard for managing infrastructure, infrastructure and applications that run inside. And the reason for Kubernetes is to enable organizations to create, deliver faster. So when we spotted this opportunity back then, we decided to try and put Postgres inside Kubernetes to see if the database could be part of this infrastructure. So the second question is why Postgres? Okay, why Postgres is, I've been using Postgres for 25 years. I'm a recognized Postgres contributor, and coming from a DevOps world and seen the potential of Kubernetes, we immediately saw that running Postgres in Kubernetes was not right back then. It was five and a half years ago. And we looked at the existing solutions back then, but our impression was that they were designed to bring Kubernetes to Postgres people rather than the other way around.

(01:44):
So there were tools targeted for a database audience rather than a Kubernetes audience, which is much larger. And that's where CloudNativePG started. The idea was to unify these two worlds, the world of Kubernetes and the world of SGEs with an operator designed to bring SGEs to Kubernetes people. The idea, initially the software was closed source. It was started in my previous company, and then it's become, it's been open sourced two and a half, almost three years ago. Initially, we applied for the CNCF sandbox. The CNCF is the Cloud Native Computing Foundation, which organizes this amazing event. And the idea was to donate the software to the CNCF. It didn't happen immediately. It took two and a half years in January. It happened, and I'm really glad that CloudNativePG is now a project of the CNCF, and this is where we stand in today.

Sean McManus, TelecomTV (03:01):
Gabriele, tell us about the real world adoption that you've seen for CloudNativePG.

Gabriele Bartolini, EDB (03:07):
Okay, so CloudNativePG, since the moment we open sourced it because of this different mindset Postgres gets for Kubernetes people, we've seen a steep adoption. If you look at the number of stars of CNPG, it is impressive. I mean, now we have in number of stars, the most popular operator for postgres, and this as by also presenting at various KubeCons. This is, I think the fifth or sixth cubicle I present over time. So with patients, people have tried to embrace the fact that they can run databases in Kubernetes. Also, with our involvement in the data on Kubernetes community, we are helping change the mindset. Another important thing is that now databases are the number one workloads in Kubernetes, which is fascinating if you think that this solution was not designed for stateful workloads. So having an open source project that is now part of the CNCF encourages adopters all over the world to share their experience with the world and with the hope that the software goes even further in the CNCF project to reduce the risk of vendor lock-in for the database and for Kubernetes, which is a major asset for their organization.

(04:42):
So in terms of adopters, I can cite for example, IBM Cloud Pak, Microsoft Azure, Akamai. They just did a talk before about this, and they talked about CloudNativePG on this, but Novo Nordisk is another adopter. So the list is long and becoming longer every month. So this is where we stand in terms of sector. We pretty much work with any sector because the database is everywhere, and you'll see AI is fueling the adoption of databases in Kubernetes. And also it's an opportunity for us to grow both in SKAs and Kubernetes.

Simon Metson, EDB (05:26):
Yesterday was data on Kubernetes day morning, and the sort amount of people that were coming up to Gabriele and the other maintainers and committers were huge. Some people were treating Leonardo as a rock star, which was kind of nice. And just the sort of general love for the project. It is such a, people need a database at the end of their application, right? You've got to store the data else. It's not really an app. And being able to, as Gabriele was saying, take the database into the Kubernetes ecosystem just resonates Really, really, it's a really important thing.

Gabriele Bartolini, EDB (05:58):
And I think we've taken the right approach. It took time. It's not been easy, but we've come from the ground with the CNCF working with Tech Storage, which is the authority, the technical advisory group that covers databases. And I think we've done a good job and we see a lot of recognition. As Simon was saying, people stop by saying Thank you for what you're doing, and so on. So it's really appreciated.

Sean McManus, TelecomTV (06:28):
Simon, tell us about your hybrid control plane and the need that meets.

Simon Metson, EDB (06:33):
Obviously as a company, we need to make some money and pay some salaries. And so we're taking the open source tech and building on top of that. And so we are looking at what problems do larger organizations have That Kube doesn't solve itself. CNP doesn't solve itself. We see sort of things, we have a lot of organizations where they have maybe a legacy environment that's running on old servers or maybe even new servers in a data center next to their Kubernetes environments, next to their cloud environment. And so we're trying to solve a problem where you don't want to be managing different things for different infrastructure. The whole point of Kube is to have one way of doing it. And so we are saying, well, how can we solve that problem for DBAs to data teams? And so we're looking at ways of saying, well, we are invested in Kubernetes, we're going to build the operator, but how do we extend that and expand that and cover more things?

(07:25):
And so the HCP hybrid control plane let's you monitor external databases as well as what's inside the Kube cluster. It lets you monitor multiple Kube clusters as well. And then we start thinking about what workloads are running. And as Gab was saying earlier, AI is a driver. People want to see more AI, want to get more value out of their data sets by using ai. And so how do we kind of join up those things with what we've already got? And so the HCP also provides a facility to sort manage AI workloads, also using other open source tech from the CNCF and also analytics workloads that is proprietary stuff that we've built on top of Postgres. And so we're trying to solve the data problem of a large enterprise in a sort of coherent and sensible single place.

Sean McManus, TelecomTV (08:18):
I see now organizations are increasingly concerned about data sovereignty. Simon, tell us how you help with that.

Simon Metson, EDB (08:26):
So there are a few things there. So there are some features in Postgres that sort of make it easier to stop prying eyes, look at your data, transparent data encryption, things like that, let you lock it down more. But the real thing is we are seeing companies say, we really want the flexibility of cloud, but we don't want to put it into a third party. We don't want to put our entire business into that third party. And for most companies, their dataset is their business. You can run a very high revenue company on a very key piece of data. And so being able to say, well, with Kubernetes, you've got the ability to have that dynamism of the cloud. You can run that on premise, and then you can now bring your with CNP and other tools, you can bring your dataset on premise and have that dynamic kind of behavior.

(09:20):
And then on top of that sort of AI workload, I don't necessarily want to send all of my data off to another third party. So being able to run LLMs and things like that locally gives you that sort of sovereign experience that my data is always within my four walls kind of thing. And I think people are kind of appreciating that more. I think some of the AI stuff has sort of set off alarm bells in some cases where they realize that their data set is being used as training for someone else's data set. And that's something that people are, I think, justifiably concerned about. And this gives you a nice way of saying, actually, I know that it's never leaving the perimeter.

Sean McManus, TelecomTV (09:56):
Gabriele, what can you add?

Gabriele Bartolini, EDB (09:57):
Yeah, I think to describe data sovereignty, I actually wrote a blog article about cloud neutrality for Postgres. So what we are at now is in a five stage evolution of pores. This is the last one, which I call cloud neutral. And it covers day two operations and it follows the DMAs era. With cloud neutral, you can bring your database in any Kubernetes distribution, whether it's as someone will say on-prem or entire public or hybrid, without changing anything. And this is something unprecedented. And covering also day two operations. So the most complex operational staff that can be delegated to the operator. And another trend that is important is also for costal optimization. If you want to protect your data, you can use bare metal machines for Postgres and provide better performance for AI workloads. So this is something that we are actually seeing at the moment a lot with many customers all over the world that are trying to get ownership again of the data.

Simon Metson, EDB (11:21):
I think people like to think that there's one solution, they really would like to have one cloud vendor, for example, but that's just not reality. For a large enough company. You're going to have a company that's acquired five other companies, and each of those five companies will have their own CSP accounts. They might have a data center they're running, they might be using different technologies. And so rationalizing that is a hard challenge. And so being able to say, at least for our databases, it's all CNP and we have one way of managing it. We have one way of thinking about it. Everyone in the team knows what that is you're saving, even if they're running on different clouds or on-prem or off-prem, you've got that kind of simplified experience that lets your team go faster. Whereas saying, we're having 15 different ways of doing Postgres, it's just not sustainable.

Gabriele Bartolini, EDB (12:11):
And I want to add on there with our company, we are helping a lot of customers move from yams to bare metal to Kubernetes, for example. And just a few days ago. And that fascinates me from a database perspective. I was helping this large organization in the Middle East to configure their bare metal machines with bare metal Kubernetes. And it's something that we have been advocating for a long time and seeing that it's now real, it means that change is coming, organization want understand the benefits and the opportunity of bringing their data inside. Kubernetes is now on bare metal with even local storage for also better cost.

Simon Metson, EDB (12:58):
Yeah, I think...

(13:00):
I did the keynote at DOK yesterday, and the point there was like, it's not when it's now these things are happening now. Companies are picking this technology up and using it. There was a conversation a few years ago which was like, why should I? But the rate of change is now, when do I do it? And I'm already doing it. And that's really cool.

Sean McManus, TelecomTV (13:20):
Gabriele, what's next for CloudNativePG?

Gabriele Bartolini, EDB (13:23):
And now that we are officially in the CNCF sandbox, we want to reach incubation. Incubation is when things get serious. And to get into incubation, you need adopters. And we also need to make our governance and our policies of the project conforming with the CNCF guidelines. So this is, I'd say the next big, it's another step for the growth of the project. In terms of features, I'd like to say that we are in a unique position now, and I was stopped by a few people here saying thank you for the work you're doing in bridging both the Kubernetes and the Postgres communities to make those products better. So we have worked with our company, with our engineers to introduce a patch in Postgres that will facilitate the deployment of extensions because Postgres is extensible in a Kubernetes environment. And we are taking advantage of a new feature of Kubernetes, which is called image volumes, to mount these artifacts in an immutable way so that it's secure because security is vital everywhere in Kubernetes on the fly.

(14:55):
So this is a feature that we will start benefiting maybe from next year as a community, but our company is willing to backport it to the previous versions to give added value to our customers. And then the plugins, we are making a pluggable. We have already made a pluggable interface for CloudNativePG, and we are hoping to get more contributions to the projects in terms of external plugins to make CloudNativePG not only a project, but an ecosystem that can live and thrive in the CNCF community by integrating with other projects. Now that we are in the CNCF, we can talk with leaders in other areas like observability, like Hotel Prometheus or identity management, like keycloak and so on, and grow together. So I think this is what we are planning for the future.

Sean McManus, TelecomTV (15:59):
Gentlemen, thank you both very much.

Simon Metson, EDB (16:00):
Thank you.

Gabriele Bartolini, EDB (16:01):
Thank you.