VMware Lightboards - VMware Enterprise PKS, 10 Feb 2018

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: VMware Enterprise PKS Overview

Description

VMware Product Manager Merlin Glynn provides an architectural overview of VMware Enterprise PKS.

VMware Enterprise PKS delivers enterprise grade Kubernetes with complete automation, backed by vSphere data center resources as well as public cloud services, to enable modern public cloud application development and delivery in the the Private Cloud.

For more information visit:
https://cloud.vmware.com/pivotal-container-service

A

Hi I'm Roland, Glen and I'm a product manager here at vm here at VMware and today, in this light board, we are gonna talk about a PKS overview. What is PKS are pivotal container service. It's a main goal that we have a PKS as we want to offer self-service of kubernetes at scale and I'm gonna use kate's is the abbreviation previous of kubernetes, and this is gonna. Give our customers the ability to have an agile development process in place, so they can rapidly deliver business goals and business value as quickly as possible.

A

So we're gonna start with the developers. You know there's a lot of personas that we built PKS for but one of the key personas the developer, and what does she use a little set of classes to describe the developer and we want to be able to offer the service to more than one developer, so we're not talking about one developer but developers or development teams or DevOps teams. So the key thing that we want to give them is kubernetes right, so we want to give them a kubernetes cluster.

A

And we want to offer the access to that cluster in a way that they are familiar with using. So we're gonna use cute cuddle to represent this in the whiteboard, but we could just as easily be talking about the communities API. So our goal is to offer these development teams access to our kubernetes clusters in a way that they're used to consuming them to deliver application value. Now, one of the ways we're going to do that is we're gonna, give them a new tool. It's gonna be a pretty simplistic tool.

A

We just call it PKS and it'll be offered as an API as well as well as a CLI, and so what this does is we'll actually be able to interact with something we're gonna call a. We call the pts API.

A

So through this API we can have one or more development teams, access and request multiple kubernetes clusters. So we'll just draw three clusters up here: put a context and multiple.

A

And we want to be able to offer that cute cuddle experience to all three of them for all our development teams, so be able to run applications on kubernetes to be able to really get business value out of this Yee. You also need to have a lot of other key services, one of those other key services that must exist as you need a place to store your images.

A

You need a registry of some sort, so we have VMware Harbor, which is an open source implementation of docker registry, as well as some other key services like Claire for CVE scanning and notary, so that we can even do image signing. So we can ensure that there's a degree of validity to our images that we're, storing and pulling and we're going to be able to offer that store and pull experience so be is something like the docker CLI, which is another tool that our development teams are already familiar with.

A

We can offer the capability to push images.

A

You can do a push or or most importantly, when we need to actually run our images. We'll need to do a pull in chart various kubernetes clusters, so pkas is offering this key harbor registry service. In addition, when we're running our applications, this is not just a place to run them in kubernetes and a place to store our images, but we're gonna need access to external services, and some of the services can expose themselves in different ways. So one of the one of the concepts that we offer with PKS will have a catalog API.

A

And this will be behind the kubernetes experience, so the developer would still be accessing all this through the cube, cuddle, API or through the cube, cuddle command lit or through the kubernetes api, so that this catalog api experience, we can actually offer something in the backend via something we call oz bappy, which is the open service broker API. We can actually offer services from key platforms like GCP or other third-party services. You know so far.

A

Applications need things like a message store or persistent data external from the clusters or some place to store analytical data or stream logging information. We can. We can expose that, via this catalog API, all through a consistent development experience through the kubernetes api through cube kettle. Now it's not just about external services, but you also need to have networking security.

A

And two of the key things that we really really want to offer our development community in our applications that are going to be running inside of kubernetes is the ability to do load, balancing and the ability to define policy.

A

Declaratively, meaning I can write a Yambol structure that says how my firewall should be deployed for a certain application or how that application should be exposed in a load. Balancer set and I can do all that through the kubernetes api and have this done for me in my infrastructure and last I need to find a way to offer storage.

A

Persistent volume claims, so we just put a PV C here. The ability in our infrastructure to be able to offer a persistent storage set for our applications running inside our kubernetes clusters, and all this has done is declared through the kubernetes api. So our developer doesn't have to learn any new tools or reach out to any other sets they're using these key things like the cube kettle command, lenay, the kubernetes api, docker CLI and our new self-service pts api. So they can request a kubernetes cluster to expand and offer all these capabilities.

A

So it's not all about the developer, although that you know these are the key services that we want to offer to give rapid and agile development. But we got to consider what happens you know what happens underneath is what happens actually providing all of these services for us with PKS. So what happens underneath this service layer, so the PKS API interacts with a component that we call Bosh and Bosh does a very important day. One capability for us. It offers a deployment mechanism.

A

So when a development, a developer, a development team makes a request that PKS API that API is going to is going to build something for us. It's going to build us a manifest, and it's going to work with Bosh to make sure that we deploy instances. For example, our kubernetes cluster here would be built with maybe one master node, we'll just use an M and a couple of worker nodes and we'd have a repetitive set for all of our other clusters that are being deployed all in a self-service manner.

A

Now it's not just about the kubernetes clusters. We talked about. For example, this harbor service will also have boss managing the instances for this harbor service as well and notice. I use the term managed instead of deploy so Bosch deploy, has all of these forests via request from either the kubernetes api or a tool that we call ops. Man.

A

And that's a tool, that's actually used by another persona, that's really important when we talk about PKS and delivering this agile service to our development teams and that other persona will use with a wrench here.

A

Another persona is the operator the platform reliability engineer. Then we use ops man to interact with boss.

A

Now remember, I said: Bosch doesn't just do deployment I mentioned management when we talked about this harbor instance. Here there's a lot of day to capabilities in Bosch and one of those is to monitor and repair.

A

So, if any of these instances that are serving these upper-level services to our pipelines and our applications suffer, a failure, get compromised need to be rebuilt, need to be patched or upgraded.

A

Bosch will actually help us do that Bosch will automatically repair in a case of an instances, unhealthy and and will offer offer us a way to do. Scaleable, patches and upgrades, as well as scaling.

A

Bosch also is a the exposed view. This PKS API will offer the capability that our developer or development teams can make a request to expand. For example, we can add a worker on to one of our clusters, or our operator can also make that determination, based on capacity analysis, what type of scale dynamic growth and collapse. What type of scale do we need an infrastructure, that's serving all of these services, so we mentioned here or I mentioned here.

A

That kubernetes is also extending out to all these other cool things that we offer to the to our applications and developers. You know this network security with these load balancers and policy. This persistent volumes and catalog ap is well. Some of that has to be provided by the infrastructure, so network and security will focus on that for right now, in addition to Bosch, another key component of PKS is Innes x, and our operator can actually interact with in a sex via is something called in a sex manager or the nsx T manager.

A

So what happens here is our integration with it with kubernetes is any time a developer needs a policy or shouldn't say a developer, but the application that the developer, the Python, is pushing, needs a security policy or needs a load balancer construct. Our integration with kubernetes is actually having that automatically created, it's being deployed as the event of the object being created in kubernetes.

A

There's, automation that will actually build fire distributed firewall policies, load, balancer, constructs virtual server, constructs HTTP rules all based on the declarative, the animal that's being pushed in when an application is being instantiated inside of kubernetes. It's a really powerful capability. In addition, we see our PVCs are persistent volume claims we're running underneath NSX is you're, gonna need an is, and in this case the is that we're gonna be talking about is vSphere an NP KS.

A

We've actually got the vSphere cloud provider a project hatchway, and this is the center's how the operator would actually interact with our vSphere infrastructure. Now in project hatchway, we offer the capability to expose vSphere storage classes.

A

So that they can then result in the creation of persistent volumes and persistent volume, claims and kubernetes, so that we can give static arts that we can get persistent storage to our application base. So we put all these things together and this this is PKS, and we will look at this, this component of the PKS API and Bosch and NS xt these components, these key components make PKS, and what PKS does is this I mentioned before our main goal with PKS is that we can offer self-service kubernetes at scale in production for your enterprise?

A

A