Jakarta EE Tech Talks, 24 Feb 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Jakarta Tech Talk - Serverless Java Event Driven Applications in the Cloud Native world

Description

February 24, 2022

Java ecosystem is vast and rich. It has been the center of innovation for years. Many people and organizations invested time and money in Java.

...but, Java is not very popular in the cloud!

Who says you cannot use Java in the cloud? Would you like to use Java in event-driven and serverless architectures?

I will show you how to do that by making a Java and Quarkus app run on the cloud using minimal resources. In my demo, I will create a simple pipeline that use Knative which provides plug&play components to process events and a Quarkus application to receive the events. We'll get the events from Apache Kafka to the pipeline and see how the Java application scales quickly and with minimum resources.

---

Jakarta EE gives developers a comprehensive set of vendor-neutral, open specifications that are used for developing modern, cloud native Java applications from the ground up. With Jakarta EE, technology developers and consumers can be confident they have the best technologies for developing cloud native, mission-critical applications. And they can build on decades of Java developer expertise to move existing workloads to the cloud.

Learn more: https://bit.ly/3IPrxOq

Follow us on Twitter: https://twitter.com/home
Follow us on LinkedIn: https://www.linkedin.com/showcase/jak...
Like us on Facebook: https://www.facebook.com/JakartaEE/

A

Hello, everyone and welcome to another jakarta tech talk. My name is samantha and joining us today. Is ali ali works on the native project for red hat, with a strong focus on event, driven aspects. He is an approver in multiple native modules, including the core eventing module he experienced with many other emerging technologies in his free time. If you have any questions for ali as we move through, today's presentation feel free to ask it in the chat or use the ask a question tab without any further delay ali over to you.

B

Thank you very much, samantha hi everyone. um Let me share my screen already.

C

Okay, um so yeah, my name is ali. I work for red hat, I'm a senior software engineer at red hat, um I'm working on k-native eventing, um so.

B

K-Native uh today, I will actually talk a lot about k-nato and.

C

How you can use it in your java workloads? Basically, I'm.

B

An apache commenter since 2010 I was uh initially working on a project called my faces. I've seen some people from the my faces community uh that who attended.

A

The the talk- and it made me actually really happy.

B

And I live in istanbul in turkey.

C

So I will talk about k native. I will also talk about corkus.

B

I mean these are really big subjects, so um I will briefly uh touch on both to give you an idea what these are, and I also have a demo.

C

um So canada.

B

Is about serverless on kubernetes, but you know how the evolution of the uh software architectures happen. So initially we had the gold good old monoliths, but they weren't very efficient. You couldn't scale them and you couldn't utilize your resources better, so we needed to break them into smaller parts and then came microservices.

C

And containers, along with it and then.

B

All these containers needed orchestration and then you needed kubernetes and eventually now we have serverless on kubernetes with k. Native serverless is already a concept that is known since many years, but outside of the kubernetes area.

B

C

I'm going to show you is serverless on kubernetes, okay. So what is.

B

Knativ, it's a kubernetes based platform to deploy and manage modern serverless workloads. That means it is running on kubernetes and it runs on any kubernetes, no vendor lock-in.

B

That means, if you have a private cloud or if you run your things workload on public cloud, you can run you can use. You can make use of k native.

C

On your workload and another good thing, is, you can still use.

B

Your uh existing kubernetes uh resources, like deployments and replica sets and whatever you have so if you switch to kubernetes, if you install kubernetes or if you install k native, you, don't have to uh switch everything, you don't have to go all the way you just need.

C

You can just use k native uh wherever it's needed and uh for, for example, this makes sense. If you have some legacy.

B

Applications that you want to um switch.

C

To migrate to k native, so you can do it step by step and canada is not just functions like the serverless. As we know it.

B

Because of basically this, so it's still leveraging all the kubernetes primitives, so not just functions. Stateless functions.

C

So k-nato has two modules. The first one is serving. It is basically to run your applications and as parts and provide scaling and networking and etc and eventing is about managing your events and basically having um something like.

B

Event driven architectures, both modules can run standalone, but it's better if you use them together.

C

So k native is a circa native serving module.

B

Provides the most known feature of it is auto scaling and it.

C

Does auto scaling with scale to zero? So whenever there's a peak in your load, it.

B

Just scales to uh even hundreds of pods, if there.

C

Is nothing it scales down to zero um and it also provides advanced networking in the networking. So it's not just about uh like scaling parts.

B

It's also about networking, as in you can create revisions, you can use traffic splitting and things like that. So that means, when you have a.

C

K native serving service.

B

You can you can do things like uh canary deployments and a b deployments and things like that green blue deployments, and things like that.

C

So just have some part of the uh some uh some.

B

Part of the requests, some portion of the request going to a.

C

Specific version of your uh workload.

B

And it provides immutable revisions, so that means uh it's very easy to roll back to a previous revision. If you have, uh if you found out, there's something wrong with the new revision new version of the native service, which is your application, so you can just do go back to the previous.

C

Revision and it will uh okay native serving well actually what it will do is.

A

B

um It will not only revert the the images for the container; it will also revert everything like the environment variables.

C

Configs and things like that,.

A

To a 2d previous revision.

B

And it has some simplifications: these could be also thought as kind of limitations, but these are done to make things simpler and also making the entire model uh more more senseful. So the thing is, uh you can only have a single part exposed.

C

So that k native serving will you know.

B

It will monitor actually that part the request coming to this uh pod on that port, and it will scale down based number based on the number of requests. You cannot have persistent volumes when I prepared.

C

Slide this slide. Actually, there were no purchasing volume support. Now there is um so. This is not an issue anymore. Of course, there are some limitations um and.

B

You can only run single containers.

C

I will also talk about k native eventing, so kennedy renting is about running your serverless workloads and with.

B

Event driven architecture, what that means is uh similar to requests that are coming to.

C

Like in k native serving.

B

The auto scaling happens based number based on number of requests like http requests. Here it.

C

Is very similar, but.

B

Here the requests are more like events, so, uh for example, you can get events from your kafka topic and then, if there are just too many events or messages uh in the kafka topic, then you.

C

Scale uh up your.

B

Application based on the load on the or.

A

Back pressure, or whatever it's called on the kafka topic.

B

So it consists of a core system, so there are some primitives like subscriptions.

C

Channels um things like that, but there.

B

Are also uh additional components um that.

C

Are plug and play that are that are designed in a plug-and-play way, uh so, for example, you can have event sources for kafka channel as well as um some other technology like I remember, there were um older channels which were even.

B

Doing things like uh websocket channel and things like that so for I will.

C

Talk about example, channels later in the slides.

C

um So these are like the building blocks, um so we have sources channels subscriptions broker, etc. I will talk about these uh in a.

B

Moment but to start with, let's take the simplest one, which is the source.

B

C

K native source does is it gets a message from a message: origin such as apache kafka, the kubernetes api server itself, aws quest.

B

Github or pretty much any container you have, so you can have your kind of legacy. Containers acting like event, sources or the source is getting, can get events from from your legacy containers and what it does is the source converts, the original origin, specific event to a cloud event. So a cloud event is something that is used in the.

C

K native there is a specification about that.

B

So um we will see examples of a cloud what what a cloud event looks like.

C

So what the source does is basically, then again uh converting the origin.

B

Specific event into a cloud when and send the event to its sync.

B

And we have actually all these sources in the k native community, we have an apache kafka source. We have kubernetes api server.

A

C

Square source github source, a bunch of other sources, for example,.

B

The apache kafka source is useful if you want to read messages from the kafka topic, and if there are too many messages, then you can scale up.

C

B

Application that is, processing uh the events from our messages from your topic.

A

B

uh We have kubernetes api server source, for example. What it does is uh using this. You can listen to the kubernetes api server events.

C

B

New part is created or a node is down, and things like that and you can get events to your application and then you can do something about it. Maybe send an alert.

C

Maybe do something else um so aws sqs is very similar to apache kafka in.

B

In a sense that it you just.

C

B

C

Or yeah messages from it and then you take the messages into the.

B

Event pipeline that you can create with k native components.

C

um Yeah github is also another example. So uh what you have what you can have here is you? Can um I don't remember if it's actually doing poll or a push um like if there is a web hook or if it's pulling uh polling either way what source does is it gets your messages like there's a pr created and or there's.

B

A new repo created some comment added something from github and then it creates it to it, converts it to your cloud event and then that way you can use this event in your pipeline.

C

So yeah, this is the very simple example. I will actually demo this one exactly um the message: origin.

B

Sends the event to source or the source pulls the event from message: origin.

C

B

It converts it cloudwhen and eventually sends it to async. So what I think is I mean.

C

In eventing or in messaging uh terminology.

B

Sync is basically something you send messages.

C

To um and here in k, nato, we have example, things uh k native service can be um it's a sync, so the k native sources like kafka source, for example,.

B

Can get messages from the topic and then it can send it to the sync.

C

The good thing is uh so these are all plug and play uh sorry, not all, but there are also plug and play key native syncs like there is a sync for uh kafka. There is a thing for redis.

C

That means um what.

B

Plugin play here means is, you don't have to write any code to send events to a kafka topic or to redis you just.

C

Use the existing things that are provided to you another example same could be a kkk native channel. I will talk about channels later uh to emphasize again on the k native service, um so um your service is basically so a k-native service is basically something you write and it's your application, it's your code. So if you just use this application canadian service uh as your sync, that means.

B

You will write your own code to process the events.

A

That are coming from the source.

C

um So before going more detail into k native because there are other stuff like channels and brokers and things.

B

Like that, I also want to talk about quarkus.

C

So quarkus is a.

B

Stack basically, it's a kubernetes native javascript that is tailored for open jdk hotspot and crawl vm crafted from the best java libraries and standards. So that means it's. It has best read of java libraries and standards. It's kubernetes native. uh It's a java stack.

C

It has lots of stuff like um it is eclipse micro profile compatible um using all the.

B

Crawl, vm and all the new uh technologies in the java ecosystem. It uses a low memory and disk and it starts really fast.

C

And yeah again with gravel.

B

Vm, uh even without crawl vm, it starts pretty fast and it uses.

C

Less disk and memory, but with crawl, vm.

B

You can create a binary and you can actually create even.

A

You can make use of even lower resource usages.

C

So a quercus application is basically your.

B

Typical maven application, you define a dependency to um to.

C

Quarkus uh artifacts here we have carcass.

B

Hibernate validator, when you define this, you just use your standard stuff.

C

B

uh Javax validation and I will also show.

C

Other stuff or the one I used in my example in my demo, is very simple: jax rs application. So.

B

This is simple: this is a simple greeting resource. It uses all the small uh all the nice stuff from micro profile and.

C

Jax rs and it's a pretty standard thing, so you don't have to. um There are some caveats, but you.

B

Don't have to write something special for quercus.

C

um So I will show a little bit more code in my demo about corkus, but so here is my demo. What I will do is I.

B

Have a message.

C

B

C

Is um so all of these is all of these are running on kubernetes? I have this message producer. It creates.

B

Events um like 20 000 messages, it writes them into a kafka cluster, so a topic and the kafka source.

C

Will get the messages from the kafka topic and it will.

A

Actually send them to the sync, which is my custom application.

C

C

And yeah this is the url for the demo.

A

A

C

C

Intellij is.

B

Showing me the yeah okay.

C

Okay, so I have the message generator here. This is not a java application. This is not really relevant.

B

Here I just want, I just needed something: uh to send a bunch of events to a kafka topic, a bunch of messages to a kafka topic, so I'll just skip that one for now. Well, what you need to know about this, is you just give this container the container and image built from this application? You just give.

C

This a url and it will send to 20 000 messages and the other thing the other custom application. I wrote is this message sync, so this is actually a quercus.

B

Application, so you can see the maven.

C

B

C

B

Is not something strange here, it's all like.

C

The stuff you know so, but they are kind of tailored.

B

um And the only code, the only endpoint. The rest end point I have here is this recent greeting resource um so again from the previous from sleds you can see. These are the standard things that I use here. Nothing strange and there is this greeting resource that is handling.

C

This path- um and this is the this- is my um like.

B

Request handler, so it's a accepting posts, it's producing json, it consumes everything and.

C

What it does is, it just gets the headers it iterates over the headers, and then it writes the body so to so. It.

B

Writes the headers and the body to this standard output.

C

um Yeah, this is really it like. I mean it's.

B

A really simple carcass application, um it starts I'll, show it. It starts really fast, which is really important. It.

C

Uses uh low uh memory and uh low disk so it.

B

Has low risk requirements which is really good, which is what we need in cloud in kubernetes.

A

So, let's build this thing.

C

Okay, I'm already here in the.

B

In the message, sync application.

A

Let me do this: oh.

C

Okay, so all the stuff.

B

You know palm and the only strange thing or not the strange, but the only uh different thing is that I have this greeting resource now to build this. I have some notes.

A

Somewhere, let me bring that up. Yeah.

C

So what I can do is.

B

I can write just run this command, uh compile and then quark, steve.

B

So this is doing the typical compile and then it will start a server and.

C

Yeah it it's uh it's running.

A

B

And let me send an event actually a sample cloud event.

A

I don't have to send the cloud in, but I just wanted to uh what it does. Is it just iterates over.

B

Headers- and it also writes the body response body to the output.

C

So this is my simple curl, which I have some headers and a.

A

Body for the request.

C

So I mean this is very simple application. What I actually want to show is more interesting stuff like what you can do with um there is.

B

Hot reloading and things like that, let me also show you that one.

A

I forgot okay, so let's run the same thing again: yeah and I'll see headers body and things like that. So let's go into the code and.

C

Change headers to headers.

A

To and then save.

A

Does it actually say yeah.

C

Yeah it didn't work because my id is not really saving anything yeah. Okay, I will skip that one sorry about that.

B

um So the next thing is what I can do with all of this mom.

A

Is sorry I'm getting okay ali? We do happen to have a question in the chat which says: how do you know it consumes low memory? Is there an observability tool built in to capture this.

C

um There is no.

B

What I can say is, I will also show you about that. I will show uh the memory of usage directly on the operating.

A

C

For simplicity, um I.

B

Think you can make use of existing.

C

uh Observability tools and you can have I'm pretty sure there is some plug-in for memory use and things like that. Parker's plug-in that exposes.

B

Memory, usage and other things that you can capture them in a central location and have a look. But.

A

I will not show you about this, but I will show what.

B

I will show, is I'll, show the memory usage directly on the operating system.

A

Here on my machine, why.

B

Blocking used at consumer code.

B

A

I don't remember.

C

I I prepared this some time ago.

A

Let's delete it, if you like and see what happens, oops.

B

Yeah nothing, so I think I copy.

C

Pasted that part and yeah, you know normally with all these new nio.

B

Systems what you can do is you can do things non-blocking.

C

But I'm not 100 sure.

A

B

So the other thing I will show you is building the native.

C

Executable for this carcass application.

B

So this takes actually some time like 30 seconds or something maybe more since I'm now sharing my screen.

C

uh What it does is um it uses the growlers coral vm.

B

To create the native binary and then you can just run the binary and you will have the same application running. So this is more useful, because this is optimized in a way that it uses less resources.

A

A

It is more efficient.

B

And it starts, it starts faster.

B

Also, if you don't have so this command, if you want to try, uh there are instructions in the in the demo readme file, um so this command is basically building the native directly on my machine, but there is.

C

Also, a there's also another command that I.

B

Put there when you run it, you don't have to install growl vm on your machine, but you just use the crawl vm in a container.

B

B

A

Okay, this is taking longer than I thought. Let me show you the command at least.

A

B

Container build so it builds the same thing, the same binary, but using a container.

A

So we don't have to install bralian.

A

Yeah, okay, finally,.

C

Finished so now, when I check target I can see there is a file called message: sync runner: it's a let's check, it's a it's, a 44 megabyte application and if you run it.

B

C

See the same thing, so it's the same quarkus application and it will produce the same output. So this is more useful uh because let's do this so I have very simple command to.

A

A

B

So what I do is I first print the date and then I run the uh application, so we can have some kind of feeling of how fast it starts. It's not the best measurement tool. I know, but I mean, if you check this time and then this time when it was ready, you can see it's like really.

A

Milliseconds and it's again yeah running.

B

And the memory usage, let me also show you that one.

A

Maybe do it here so.

B

This is the ps command with which shows the rss the uh yeah. These are complicated concepts again. This is not.

C

Like 100 accurate, but what my.

B

Ps command shows me is this: application is using 18 megabytes of ram, which is.

A

Something when you think about a java application.

C

B

So I just showed you about the um the part about like how to.

C

Build your carcass application, and things like that and.

B

It's running really fast, it's using less memory and things like that. But now.

A

Let's actually run this application in a container.

B

So in my demo I have these folders.

C

B

Is this, as I showed you message generator, which is a node application, not relevant message, sync is the just the application that I just built. I.

C

Also built an image for that and I pushed it to uh to the docker hub so.

B

I will just use that and if you check the config folder.

C

There are files like these, so the first one is actually just creating the namespace. The second one is creating the sync. The third one is creating topic, so the namespace is yeah. The namespace.

B

The sync is actually my message: sync.

C

Application here so the second.

A

C

Will create this one um yeah? Let me maybe first run them.

A

So I just create the namespace and then.

C

The sync: when I create this thing, I will see.

B

Some pods on my namespace because.

C

I created basically a k native service and when.

B

You create a canadian service, it will start running immediately uh it doesn't. It won't wait until there is load or something, but after some time I don't know how much maybe 60 seconds or something uh if there's no load it will I mean, does that is the default uh timeout?

B

If there is no load, then it will scale down the deployment.

A

So the pots the service, and then there is a topic, a kafka topic.

B

That I will send messages to.

C

And then there is this source.

B

If you check the source, this is the source, so it gets the events from kafka cluster and then it will send it to.

C

The sing when I apply this, the only piece that I didn't apply that I didn't create basically will be the message producer. So I will just do that.

A

As a last step,.

B

Okay, so my kafka source pod is created, so the kafka source pod is running always.

C

To you know to check if there are any messages, but my sync, my k, native service is scaled down because there were no uh requests.

B

And let me do some additional things here, so I will use this turn command, which is basically uh to it's following the logs for the sync application.

A

Yeah so shut down complete exiting.

C

Yeah there are no k native service pods.

B

Anymore for my application- and I will also start- I will also follow the logs for message generator, which.

A

I will just create.

C

So when I hit enter.

C

B

Message generator starts sending 20 000 messages to to the kafka topic, and here you can see my k native service is scaled up to.

A

I don't know 10 yeah 10.

B

C

Because there are too many messages, there are too many requests from the kafka source to this thing it actually the kaikai native, serving scales, this service up to 10 parts and now messages are all sent, and these are my logs again for the sync application. Different sync pods, um like I mean they're like intercepting like they're mixed, because there are different containers there. There are different parts here again, the same applications.

B

That I just showed you body.

C

And headers, so if you wait like 60 seconds, we.

B

Will start seeing these going down.

A

Yeah they're already being terminated.

A

So that was basically it about the demo.

C

um So I'll share.

B

The link for the presentation where you can find the demo uh link to the demo here.

C

um Okay, so again, this is the source to sing scenario, and there are also other um constructs.

B

In k, native eventing, one of them is a channel channel is a bit uh I mean channel is different than.

C

Source, what channel does is it lives in.

B

And again, in the k native eventing system, uh when you send events to it, it provides fan out and basically something like pops up.

B

um So when you send these two events, one and two to the channel- uh and you also need to create subscriptions like subscription uh resources that defines something like this- I am subscribed to these events on this channel um and please send the events to this sink. So what channel will do is it will send these events? To these things? Example channels are in memory channels which are not really, which is they shouldn't be used in in production, loads.

C

But they're useful for testing stuff and.

B

Again, apache kafka channel. So when you send an event to a apache kafka channel, the event is sent to apache kafka first and then it will retrieve again from apache kafka. What this.

C

Provides is um persistence first of all, so if your channel or if your sinks or something.

B

In between are down, if they crash, you will still.

C

Have your events there and the second part.

B

Is if you are producing way faster than you consume, using using a channel will help you synchronize.

A

B

A

C

Your your processing.

B

uh What I mean is, uh if you send thousand messages to your kafka topic in a second and if you process only 100 a second in your sinks. What you can do is you can scale up, but if you're using, uh if you decide you're using too many resources, then you can.

C

Limit the scale up, that's also a feature in k native serving, for example, um and you can say like I.

B

Want to only use five uh uh five replicas, then there will be a huge difference between your producing speed and uh consuming speed. Like hundreds.

C

uh Sorry thousand messages per second versus five uh 500 messages per second, um so normally.

B

You cannot your application typically, if you send them directly, they won't, they cannot handle it, but what channel does.

C

Is you send just you just send the messages to channel and channel um works in a way like it makes sure that.

B

The sync process, the messages it sends so it just sounds like sequentially.

C

um So this helps with the synchronization um yeah. Another example channel is the gcp google.

A

C

Yeah so, for example, this is called publish, subscribe.

B

Channel in the book called enterprise integration patterns, which is a kind of unknown book for doing messaging and integration, and things like that. So what how you can do this with k native is using channels.

C

Like um if you have.

B

Here we have two different message: origins, the one. The first message.

C

Origin is outside your kubernetes cluster and the second one is living.

B

Inside your kubernetes cluster, this could be your legacy application or something- and this could be.

C

Your message, origin, uh sorry, your your external service um and.

B

These sends events to your channel uh to make it with the external message origin. uh It cannot send anyone to channel directly, so you would need the source to convert that event to a cloud when, but if you're in.

C

Cluster application or a another k-native source that is living in your kubernetes cluster.

B

C

B

Is producing a cloud winds? It can directly talk to the channel, so channel collects all these messages and.

C

Delivers them to different things by subscription.

C

B

Okay, the um next one, the next kind of primitive in canada or the building block. I should say in.

C

Canada eventing is called broker broker is pretty much uh it's very similar to a channel, except it provides a more.

B

Advanced filtering, uh so what it means is with these filtering, you can.

C

Think of this broker, like a kind of message, bus or a regular broker, that.

B

You can rely on sending events and loosely coupling or.

C

Decoupling things um so.

B

Again, if you check the book enterprise entry integration patterns book, this is kind of mentioned. As content based router, you have different sources here. They send different kind of events so event screen. Events will do like first, one could be a github event and the second one could be a kafka event event, so the broker collects them, and then here very similar to subscriptions before you create triggers, and in triggers uh you define what kind of events you like to.

A

Get delivered to the sinks you define.

C

B

With this trigger, I say we only need. We only want the green ones like github events, and then this thing will only get github events.

C

Similar story here.

C

um So there are also other things like plugins, so these are like plug-and-play components. What I mentioned, and there are also other plug-and-play components.

B

Like syncs and there's also something called sequence, a parallel sequence is basically to pipe to.

C

uh Services, for example, okay native event, sort of circadian serving services, so that the first.

B

One produces something and it becomes an input to the next one and you don't your application.

C

B

Like in your code or somewhere, you don't have to define the url of the second one, so it's providing some kind of bounding or binding um parallel is similar. So when you send something to a parallel, it will make sure these are sent to different uh channels. So it's again only fan out um so one example thing that I will show you, because I will. I hear a lot about ml ml and.

C

B

You can do with k native, so this is like a typical ml pipeline. I'm not an expert in this. I just know: there's a collection phase, you receive you get events or data, basically from different clients. It could be twitter, it could be your iot device, it could be whatever, and then you ingest these data, you store them somewhere.

C

And then you prepare these data like.

B

You remove some stuff, you make them, comply to a common format or something like that, and then you do the computation on them and then eventually, when you do computation, you store the final result, result somewhere and then you can do a presentation like you can show a graph. You can send an email. You.

C

Can send an alert.

B

If there is something that needs to be taken care of so this is like a typical pipeline, and this could be how you can do this with k native.

C

B

Here I have three clients. The first one is talking to kafka directly. Excuse me.

C

And the second.

B

One is talking, uh rests talking to a rest, endpoint and the third one is talking mqtt, these two rest and mqtt endpoints. They can also send.

C

Their um messages to a kafka topic.

B

And then, with kafka source, you can get the messages.

C

Or events from the kafka topic and then you can send them to a k-native sequence, where you do the data preparation.

B

Like the k-native service one, it can do something. Maybe it can remove some fields or you know it can apply some kind of transform.

C

And the second service, it can do another kind of transform.

B

And then, eventually, the output of the sequence could be sent to kafka using a kafka sync and then from kafka.

C

You can have your presentation with like dashboard or you can send emails. Maybe there is that you could use another application here, which is not really relevant. For this example.

B

But the good point is this: so if I switch to the next slide, you can see everything stays the same like the applications, the code, I wrote, for example,.

C

For clients or for this transformation applications they are the same. So if I.

B

Switch from if I want to switch from kafka to google.

C

Cloud slot, storage or some you know some other technology, um then what I would use here is instead of uh kafka. I would send my.

B

Data to google cloud storage and then there's a source for that which I can make use of and then eventually.

C

I think I don't know if there is actually a plug-in placing for google cloud slot storage, but you can use one of the things or you can.

B

Just write a small application here which sends events to the target system you want. So this is what I mean by plug-in play components. We already have kafka source. We already have k-nato sequence. We already have kafka singh, so you will not write any code for these. You will only write your own uh own logic and there is no vendor log in, as in you, can run k native on any kubernetes distribution and also there is no vendor lock-in when it comes to like integration. So if you use kafka here, you can also.

C

Make use of other uh plug-and-play components or other um messaging systems or any other services.

C

So these are my kind of takeaways canadian provides better workload management.

C

There are also a lot of features, especially in serving um like these.

B

C

B

I mentioned about traffic splitting doing uh experimental deployments and things like that.

B

So all these good stuff, on top of being able to scale dynamically, including.

C

Zero um and.

B

Quarkus is leveraging the existing java ecosystem with all these standards and with compliance to a bunch of.

C

Standards- and it is actually pretty good for.

B

Serverless, what we need for serverless it is.

C

B

C

B

C

It's starting up real real fast, it's using very minimal.

A

B

So that's it and you can reach me out on twitter at this, handle.

C

A

If you have any questions, I would be happy to want to answer.

A