Jakarta EE EclipseCon 2020, 26 Oct 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Evolving Jakarta EE Applications in Kubernetes

Description

The session will describe how to migrate an existing Java EE/Jakarta EE application to run in Kubernetes, how that application can be extended using Java microservices supporting MicroProfile. I will show how a "hybrid" application comprising Java EE/Jakarta EE components, microservices, and a shared distributed cache can be modeled and managed as a single logical application in Kubernetes to support ongoing application modernization. The session will demonstrate technologies that make such a migration possible, giving developers pragmatic approaches for evolving existing enterprise applications to modern cloud native architectures and innovative technologies such as GraalVM.

Attendees should be generally familiar with Java EE/Jakarta EE technologies, MicroProfile and containers and Kubernetes/CNCF technologies. The session focuses on how these technologies can be integrated and used together to support evolution of cloud native enterprise applications in Java.

Speaker(s):
Will Lyons (Oracle)

A

Hello, everybody, my name is, will lyons and um I manage product management for oracle enterprise cloud native java and um I'd like to talk to you today about evolving jakarta ee applications in kubernetes.

A

um I've been participating in the jakarta e working group with the eclipse foundation and it's very important to us that we evolve jakarta ee to support cloud native applications, and I'm going to talk today about how you can do that and how you can do that, specifically to run jaccardi applications in kubernetes and cloud native technologies and how you evolve them most of the information I'm going to be talking about is available today. Some of it is forward-looking and that is covered by this oracle.

A

Safe, harbor statement.

A

Okay, I assume most of you are familiar with jacardi e, but I will just review it briefly. Java enterprise edition has been proven for many years. It was a set of technologies that was layered on a java standard edition.

A

It continues to be supported. There are seven or eight now versions of java ee.

A

Several years ago, oracle made the decision that it wanted to donate the joby technologies to the eclipse foundation.

A

We formed a community, a working group at the eclipse foundation and contributed the technologies and the specifications our implementation of glassfish to the jakarta ee working group at the eclipse foundation, so jacardi ee is now the represents the evolution of java enterprise edition technologies.

A

We have a release that's available today.

A

Excuse me, I had a zoom interruption there. We have a release. That's available today called jakarta ee8, so jacardi e represents the evolution of java enterprise edition.

A

It's proven broadly in many enterprise applications. So around the world.

A

Most many large companies are running java, ee or jakarta ee applications in production, jakarta e implementations have proven their value in terms of delivering applications that run at scale and deliver deliver very high levels of performance that are robust and deliver high availability in the event of failures and underlying infrastructure.

A

The dracardi implementations typically provide high availability to end users and reliability in the data that's being used in those applications.

A

Jakarta e implementations uh have evolved to provide very sophisticated management capabilities to simplify the process of managing applications in production and, generally speaking, the uh implementations are proven. They're secure, they're integrated with the master array of technologies, databases, web servers, other implementations that are built on top of them. So it's a proven platform.

A

However, in the industry there's lots of change occurring.

A

We see broad interest in uh moving to cloud native platforms uh where applications are deployed in containers they and that are deployed in containers that can be easily moved across public and private clouds, interest in leveraging devops technologies for automation of delivery of applications through test and production, and the ability to monitor those applications in production uh interest in building out microservices, architectures uh and serverless applications not only implemented in java but implemented in many languages, and so we are hearing- and I am hearing from many customers- uh who've implemented applications on java, ee or jakarta e.

A

Now, how do I combine the best of both worlds?

A

How do I take my existing applications, which are managing mission critical workloads, and how do I monitor the modernize them so that I can take advantage of these new cloud native technologies run my existing mission, critical applications in these new technologies and infrastructures leverage some of the benefits that docker and kubernetes provide for these applications that I have enabled me to easily move my applications across my own private clouds or public clouds, and enable me to evolve those applications either running in to cartier implementations or evolve them towards microservices technologies.

A

So that's a question we are hearing frequently from our customers and to state it another way. The question is: how can I containerize well: I've got an existing portfolio of applications running typically on physical machines and virtual machines on premises.

A

How do I um containerize those technologies, run them in kubernetes, either in private clouds or in public clouds and evolve those technologies there or gradually adopt microservices technologies like micro profile, to deploy applications that can coexist with my existing uh jakarta ee applications and as I evolve my application environment. How can I manage my existing portfolio of jaccardi applications alongside these net new microservices apps that I'm developing and what we're hearing from our customers? How do we do this, which does not in a manner that does not necessarily require that? I immediately rewrite it everything?

A

uh How can I evolve at the pace that I choose with the applications that I choose uh in a time frame? That makes sense for me.

A

So that's the question, we're hearing a lot and we've done a lot of work with an oracle to enable that and one key piece uh or an enabling technology is kubernetes operators.

A

I'll review this briefly for in for those of you who are not familiar. A kubernetes operator is a standard kubernetes concept.

A

An operator is essentially an extension to standard kubernetes to manage uh application, specific types of applications and the components of those applications and, in this case, we're talking about an application type, which is an application deployed in a jakarta e implementation.

A

The basic notion is an operator leverages. The notion of a custom resource in kubernetes in a kubernetes customer resource enables you to define a resource and state the way you want that resource to be running in kubernetes and an operator is a controller that is monitoring the state of the kubernetes cluster, assessing that versus the what is defined in the custom resource and is constantly monitoring and modifying the actual runtime environment to meet the definition that's established within the custom resource.

A

So if the custom resource says, I want uh five pods running this application with these characteristics, the operator is constantly monitoring the environment, to ensure that you do in fact have pi five pods with those attributes running in your cluster.

A

It turns out that an operator is an ideal technology for automating life cycle management of jakarta e applications.

A

So I'm going to do a little demonstration of this uh and in this demonstration I'm going to be showing the kubernetes operator that has been implemented to manage weblogic server applications.

A

So we've uh implemented an operator using a standard kubernetes pattern. It's implemented in open source we've got production deployments, we're building services in oracle cloud which leverages tech this technology.

A

Other products from oracle that are built on weblogic server are leveraging this kubernetes capability.

A

Applications built by oracle are also deployed and using this kubernetes operator, and this kubernetes operator we built for weblogic, actually represents the core of a set of tools we've provided for managing weblogic applications running in kubernetes.

A

So let's go to my demo. I hope you can see my screen so um here I'm showing um a screen from the oracle cloud infrastructure console- um and this is within within that console- I'm able to monitor kubernetes clusters, which are deployed within oracle cloud and my oracle cloud identity domain, and here I'm showing a node pool.

A

So within my kubernetes cluster, I've got a set of I've got a node pool which is configured and it is spread across a number of availability domains running in our phoenix region.

A

I have three nodes configured in this node pool and they are all up and running.

A

um So, let's take a look uh at what's running in this kubernetes cluster, so first we can take a look at uh the name space at the various name, spaces or virtualized environments within uh this kubernetes cluster and we've in this demo. Environment we've created a variety of namespaces, in which we can demonstrate various uh applications that are running and there's actually three that are going to be relevant for this demo.

A

We've got a sample domain namespace, another namespace in which the weblogic operator is just referring to runs and another namespace in which our traffic load balancer is running. So, let's uh take a look at the.

A

What's running in the operator.

A

A

A

So I'm looking within the weblogic operator, namespace and type this wrong excuse.

A

A

And you can see that within the operator namespace we have the weblogic operator running. So the operator is monitoring um a custom resource.

A

That is defined in we'll see the values that are supplied with the operator and the operator is actually monitoring the sample domain name, space for a custom resource, that's been created there and when that custom resource is created, the operator will automatically create the resources that are defined within that namespace.

A

So right now we have not actually defined a custom resource. So if we look uh in this in this in this namespace that the operator will be controlling, there are actually no resources defined there.

A

um We have defined a.

A

A yaml file which can be used for creating a resource that the weblogic operator will use to create the resources necessary to run a weblogic application in kubernetes, and this is the yaml file you can see. Basically, this custom resource will essentially state that a resource of type domain will be need to be created. This will actually create a weblogic configuration.

A

um There are certain specifications like where to grab the docker image that will be used to run the domain uh where the domain configuration is located, java options that are used for starting up the servers that will run in the domain and how many servers will be running in the cluster.

A

So if we, if we apply this yaml file,.

A

Through kubernetes, then the custom resource will be created and that will cause the operator to detect the presence of the custom resource and to create the necessary resources to instantiate the weblogic domain.

A

So now, if we look at the resources that are running in the domain name, space.

A

No resources found: let's try that again.

A

Why would that be.

A

A

A

Well, this is awkward for some reason. I'm not.

A

I'm not able to see, oh, I think I perhaps know what maybe I haven't created a domain secret.

A

A

A

ah Excuse me so I hadn't created a domain secret which is required to actually start up the servers in the domain. um So now, if we look at the resources that are being created, we should see.

A

Well, for some reason, the demo gods are not with me today. I must move on.

A

What we should have seen was a a weblogic domain being created, in which you can run your weblogic server applications and having deployed a weblogic application in that manner.

A

You can leverage all the tools associated with devops running in kubernetes to automate the process of updating, domains, patching domains, monitoring, domains in production and so on, and we've seen many customers successfully deploy with this technology and indicate that they are able to version their applications to patch their environments, much more simply at lower costs using this technology.

A

um So now, let's say: you've got weblogic applications running in kubernetes and you want to evolve them towards microservices.

A

You can use technologies like microprofile to do so. Microprofile just defines a set of standards which extends to cartier technologies, to enable you to build cloud-native microservices.

A

Oracle has led a project called helidon which has delivered a microprofile implementation. It's standards-based uh implemented in open source. It supports the latest micro profile, 3.3 standard, um and so this will be familiar to jacardi developers and can enable your development teams to extend uh existing applications, evolve them or migrate, existing applications uh to microprofile.

A

So this is a useful technology that you can use to extend your cardi applications. It also supports crawl, bm native image.

A

Graul vm is an open source project driven by oracle, which provides a polyglot virtual machine for running applications. It supports java.

A

It enables you to use uh graviem as your jvm.

A

It also supports native image capability, which enables you to comp to ahead of time, compilation to create a native executable, which represents your application, so you can build applications in java, compile them down to a native executable, where those applications can start very rapidly and consume very little memory, so holodon applications running on graubem native image can start up in about 20 to 30 seconds can scale very rapidly. It essentially gives you near serverless performance for standard java applications built using standard micro profile api. So it's a very powerful technology.

A

Microservices applications uh are typically implemented as stateless applications, so, since they're stateless, they are often data-hungry. uh We've seen uh customers deploying microservices applications at scale, with thousands of microservices making doing tens of thousands of transactions per second against a data store.

A

A great technology to use for such a data store is an in-memory data grid such as that implemented by coherence.

A

Coherence is an oracle technology which has been proven in mission critical applications for many years. It provides not only a cache that can be used by application, servers and microservices, but can be used for in-memory processing on the grid. We've open source uh coherence and have delivered an open source edition of coherence, called coherence, community edition it provides everything that you would need to support microservices deployment.

A

So this is a technology that can be used for your forward-looking microservices, architectures, we've also implemented or we've begun the implementation and have actually delivered a project in open source called verrazano.

A

Verrazano is targeted at enabling the combined management of hybrid applications comprising traditional web logic or coherence. Applications that leverage jakarta ee, with net new java, microservices and potentially combined with polyglot microservices implemented in other languages, verrazano leverages standard, open source tools to enable you to manage applications which span all of the type all of the application and microsoft services technologies.

A

I've just mentioned the general notion within verrazano, is you define an application model uh and your application model may comprise java ee applications, jakarta e applications, microservices built in halodon or other microservices technologies, and it may comprise data grids. You define this application model and how that application is bound to underlying kubernetes clusters, whether those be running in public cloud or private cloud environments, we provide and will provision out of the box. A monitoring stack that can be used for managing your hybrid applications and monitoring using standard tools like grafana, prometheus or elastic stack.

A

Verrazano is an open source project. That's available. Today. We've got a demo, I'm running I'll. Actually I'll give the demo a little bit of a.

A

A

This is this shows the verrazano project, which is uh implemented in open source.

A

You should instructions on how to install verizono we've demonstrated. We have some demonstration applications that you can use to test out. Verrazano functionality.

A

This is a description of the application itself, which you can deploy for testing purposes. Here is an illustration of an application model, so here's a yaml file which defines this application, which is built both using web logic, configurations.

A

It also includes microservices that are built using helidon and the combined web logic and helidon applications use a coherence cluster as their data store.

A

You also that's the application model. We were just looking at uh using verrazano. You also defined how that application uh will be bound to the underlying environment and in, for example, this yaml file describes how the placement of the application we've just been discussing will be deployed or how that will be placed on kubernetes clusters. In this case, it will be a run in a local kubernetes cluster.

A

These next screens just show the application running here, I'm actually showing the verrazano console the ui that you can be. You can use to visualize this hybrid application that is running in kubernetes and the different components of the application, including weblogic components, helidon microservices and coherence.

A

Clusters within the console, you can also view the binding and you can see where the various components of the application are deployed in which cluster, which kubernetes cluster those components are running.

A

And we provision out of the box standard grafana dashboards for monitoring the various components of the application that you have defined in your logical application model. So both web logic, helidon and coherence components can all be monitored. We provide an out-of-the-box logging stack which can be used for viewing and analyzing your logs.

A

So that's a brief run through of verrazano.

A

I hope what I've communicated in this presentation that you can take your jacardi applications and run them in kubernetes. You can deploy those applications along net new microservices and there are tools for managing those combined environments as you evolve them over time.

A

So I'm going to stop sharing.

A

Let's see if I have any questions.

A

Three questions in chat is weblogic going to support micro profile. uh It's not our plan at this at this time to for weblogic itself to support microprofile. Our intent is that we will support the development of micro profile applications through helodon, which weblogic versions are supported by the weblogic operator. Both ee, 7 and ee8 versions are supported, so that's weblogic, 1221 and 1411..

A

Well, halidon support graphql, as, for example, quarkus does um I don't know what our roadmap is on that specifically, but we'll provide an update on the helidon project to address that question, I think I'm all done in my time.

A

Thank you very much and please check out the open source sites that I've referenced and take a look at the technologies we've been discussing thanks very much.