.Net Foundation On .NET, 23 Apr 2018

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Azure Service Bus and the .NET Standard SDK

Description

Azure Service bus is a hosted messaging service inside of Microsoft Azure that provides reliable delivery between two or more parties whether they are in the cloud or on-premise. Service Bus implements the AMQP v1 standard, and enables asynchronous, decoupled communication between producers and consumers.

In this episode, Clemens Vasters(@clemensv) and Christian Wolf come on to show us the new .NET Standard client for Service Bus. Along the way, we also discuss other topics such as AMQP support, working with queues, and how Service Bus compares to other messaging services in Azure.

01:21 - What is Azure Service Bus
03:26 - Reviewing messaging services in Azure
07:43 - Mixing messaging services together
09:19 - AMQP support
12:28 - .NET Standard support
18:44 - Managing the QueueClient lifetime
20:18 - Service Bus Standard vs Premium
24:46 - Features and Integration
31:44 - Samples Repo

Take a look through the links below for more information on:
•Azure Service Bus messaging: https://docs.microsoft.com/en-us/azure/service-bus-messaging/?WT.mc_id=ondotnet-channel9-cephilli
•Azure Service Bus Geo-disaster recovery: https://docs.microsoft.com/en-us/azure/service-bus-messaging/service-bus-geo-dr?WT.mc_id=ondotnet-channel9-cephilli
•Azure Service Bus samples repo: https://github.com/Azure/azure-service-bus
•The famous issue 65: https://blogs.msdn.microsoft.com/servicebus/?WT.mc_id=ondotnet-channel9-cephilli
•Azure Service Bus blog: https://blogs.msdn.microsoft.com/servicebus/?WT.mc_id=ondotnet-channel9-cephilli
•AMQP Website: https://www.amqp.org/
•Apache Qpid: https://qpid.apache.org/
Create a Free Account (Azure): https://aka.ms/azft-net

A

In this episode of the on .net show,, I'm going to have clemens and christian joining us to talk us all about all the things that are happening inside of azure service bus.

A

Welcome to another episode of on .net show., my name is cecil phillip, and I have two special guests that are joining me, today., so today,. I have clemens vasters, and I also have christian wolf.. How are you guys doing? Today? good. fabulous. doing great., I'm so glad to have you guys, on.

B

Yes., thank you for having us. yes.. Thank you for having us.. Why don't you tell our audience a little bit about.

A

Yourselves and what do you do in your respective roles?

A

okay, I'll go first.

C

My name is clemens., I'm an architect in the messaging team.

C

The messaging team owns azure event, grids,.

C

Azure events, hubs, azure service bus and azure relay.

C

And in my role, I'm doing sanitization work on amqp on cold events.

C

And I'm also kind of the product architect, if you will, for all those services.

A

I'm christian wolf.

B

I'm the product owner of service bus., I own it from a business and a feature point of view.

B

so,. When we talk a little about what services bus,.

A

I know that's probably one of the oldest features that we've had in azure,. Probably one of the features we've had since inception, right? yes.

C

Since it started. absolutely. so,, why don't you tell us a little about.

A

What is service bus exactly, like? What do we use it? For,? What are some of the use cases? We might have that in our applications?.

B

Yes. well, service bus is an enterprise message broker., so basically we have,. I would say,.

B

Four different kinds of messaging systems which rely and talk about touch, base a little bit about.

B

And service bus is really used.

B

For business scenario, where you need high availability,.

B

A lot of security reliability,, you need anything like sessions,.

B

First in and first out ordering of messaging of messages,.

B

All you need is a feature like transactions.

B

so,. Whenever you need like a guarantee that the data that you send from left to right really arrives, and that's when you want to use service bus.

C

Yes. and the features that it has.

C

Is that it has queues, and it has topics.

C

So it gives you the basic ability to go and push data into intermediary effectively.

C

Right. then, with that,, the couple,, the sender and the receiver from another., so that, unlike with what you would do with http,, you push a message to a receiver and receiver needs to be.

C

Instantly ready to go and handle it, and service bus can sits, put that queue in the middle.

C

and with topics,. You get a pop up feature where you can go and send messages, and then only those parties were interested in that particular message will go and get it..

C

We have two, and you can also see that on the slide, here,.

C

There's two queue services and we often.

C

Get asked: why does azure has two queues.

C

Yes. and that differentiation exists, because there are cases where you have just simply that decoupling.

C

And that's the only thing you care about., you just want to go and throw tasks into a queue and want to hold the task off and handle them,. And then you have other cases. We need to have higher reliability, higher availability and need to have transactions,.

C

You need to have all the features that christian, just mentioned.

B

So service bus is specializing on.

C

Kind of the heavy-duty messaging pieces..

C

So if you look at.

B

These four systems and just touch base on each of them really quickly,, storage queues. That's exactly as clemens described like a simple task queue scenario.!

B

You want one process to run on 10,000. Maybe you need 10,000 queues for that..

B

You push everything into your queues and then the vms received from that queues and process. Whatever they have to do.

B

service bus is really for example,.

B

You have a retailer. You want to send all your post transactions orders into the cloud,.

B

You want to process them in one way or the other,.

B

You maybe want to send some to an erp system for fulfillment or for replenishment of the retail store.

B

So there's multiple different scenarios: you could think of. service bus can also be used for web shops.

B

for example,. You will sent all the orders from the web shops, process. Them, modify the messages, maybe on the go.

B

And then have actually a little sample for that that at the very end.

B

yes,, so service bus is really like bringing a highly reliable, transport.

B

and then event hubs. That is really big data streaming for analytics..

B

You have telemetry systems, you have production machinery which send signals, or you.

B

Have web pages or vms which send signals? And you want to ingest this data and start reasoning about the data.

B

And then grid,, that's the newest,. I would say kid on the block. that is for cross cloud. Reactive, programming.,.

B

It's like an eventing service., so essentially, whenever you have a state change somewhere or you have something happening, and you want someone else to react to it, that's when you want to think of grid.

B

so, typically,, the very core screen from the top,.

C

The way you think about the differentiation between the eventing services and the messaging services is that,.

C

When you use a messaging service, you typically have an intent to send a message to have someone else do something. and with an eventing service, you typically have a fact, so something just happens and you report that out.

C

So you don't have an expectation for something, so you don't need to have that degree of control, reliability and making sure that's exactly once it start happening, exactly one et cetera..

C

You don't have those concerns with the eventing and the eventing also kind of gives you some different delivery patterns that we have like,.

C

Control your own cursor with the event hubs or we do push mode with you.

C

right,. So I know some of.

A

The time the way I think about it is when I want to pass events, I'm usually thinking of something that's very, like somebody, that's I'm having a lot of throughput,.

A

Not really passing a lot of data like very small packets coming through very quickly.

A

yes., and I also think about messaging. I have like some data that I need to encapsulates and send over. I know there might be some deferred communication, right?, so like whether the consumer is available or not. You can send this message.

A

C

Whenever those consumers are up,.

A

They can now reach into it,. Now that doesn't stop my clients on the other side.

A

correct. consumers are not available. correct.. I can send a message and I can keep on doing what I'm doing.. I kind of like this asynchronous deferred type communication going on.

C

So we had a little episode from service bus,.

C

When the game halo, four launched on that day.

C

On launch day,, the entire back-end went down.

C

oh!, no data was lost because it all went.

C

Into service bus at the time and then event hubs later for the next version of halo.

C

and then they could go and process the data out,.

C

So it was a complete decoupling.

C

Of the games that were played and the back statistics back-end, that was at the end., which kind of shows you at giant scale of a game with a million players or more,.

C

That actually helps to keep that system more robust, because the gaming experience was not affected by the system being down for a while.

C

And clemens mention actually.

B

A perfect event hub scenario. all he just said that I just want to highlight that fact, that was in the past service bus is obviously our oldest service..

B

We try to do everything with service bus now that would be exactly a telemetry scenario.

B

correct., a million gamers running their games emitting telemetry, and then you have some point in time. You want to reason about the data and that's a classic for event: hubs.

B

exactly., the difference between the event, hubs and event grid is, on event. Grid,. You want to act on the data.

B

on event, hubs,. You want reason about the data.

A

So we spoke a little bit about the differences between the services, but I'm guessing we've seen patterns where we've used, multiple of them together in the same application, right?

A

C

Is that a common use case that you guys have seen?.

A

C

And we're now actually building these things out so.

C

That you can go and mix and match them very actively.

C

Yes. so specifically also with event grid,, so the case exists.

C

Very often that you have queues that are not very busy,.

A

But you need them for important events.

C

So you have a queue,, the queue receives all the alerts and those alerts should only ever happen and they're important because you must process them, so you put them into queue.

C

But those alerts only show up when something very bad happened in the system,, so ideally never but then probably like once a day or once a month..

C

So it might be very expensive to run that handler all the time. yes.. So what we can now do with a combination of event grid and the service bus, queues.

C

service bus queues can now go and raise the grid. Events that says, "this queue has a message, but nobody is listening on it.". right. and then you can go and listen to that grid. Event..

C

You turn around,, you go to the storage queue and pull that event out, but then process that with the robustness that service bus gives you. so we're starting to combine those things. But we very frequently see that customers who need messaging in the application, I've understood that gives them an architectural advantage., we'll go and look at all three services, or even the relay sometimes and combine them as they need them, because they will typically have all those various nuances of use cases in their solutions.

A

So, with these different services, different messaging service that we have,, so I know this is protocol called amqp.

A

right.. Is that something that's common across all these different services? Is that something that I could use and something that we support? Today.

A

yes,, so we use amqp for service bus and for event, hubs.

C

We also have a path how we can pass data now, to that event, hubs and to our queues using.

C

Amqp. and amqp is kind of economical protocol.

C

yes. amqp is protocol that was finalized in 2012 by industry. Consortium, is something that has been started actually in the financial industry in wall street..

C

We then eventually joined that effort, and it's now.

C

The dominant messaging protocol in the market,.

C

I'm in the technical committee at oasis for amqp,, I co-chair that.

C

And we also develop that further is actually quite a bit of activity in the committee to kind of push that forward and enable more and more use cases.

C

One thing that we're particularly working on is.

C

A lot of enterprise scenarios are driven by gms,.

C

Gm java is very broadly used in the enterprise and they have a messaging center gms, and there's a mapping for gms has been developed now in oasis..

C

They were actively participating in that and we're also helping. we're done basic plumbing all the necessary things into service bus to make that really work really well with all the features that are in there.

C

so,, that's the foundation that we have. and also to accommodate amqp,.

C

We made recently some good set of changes.

C

And so, we created a new client for service bus or a set of new clients for service bus,.

C

And including and use for client .net standard.

C

Okay, great. so, one thing that I wanted to ask.

A

You is if we're supporting amqp,, so that means that any amqp clients, right?.

A

Whether it's made by microsoft or not, should still be able to work with service bus and integrate with our services? absolutely.

C

That is correct., that's the goal. so, you can use today, and there is typically in most of the stacks now,.

C

And most of the amqp stacks now,, there are examples for service bus, specifically., sure. so,. You could pick up proton c, which is part of.

C

The apache qpid proton project and do that with.

C

C. you can go basically any language, that's supported in the proton project. You can use.

A

C

And these protocol cores that you can go and send or receive messages, with. and now,. What we're doing is we're picking up these stacks that we're also helping with.

C

Yes. and we're wrapping them to put a bit of a nicer api around them, and to make a unified api that looks the same.

C

so,. We are picking that up. For, we've picked that up for java. so,, our java,, our native java client, which is now.

C

Jms wraps the proton j clients.

C

We have our native amqp stacks for .net and .net core that are then used by others,.

C

For instance, by red hat, as the primary client for their products.

C

got it. so,. It's something that it's a give and take., of course, the industry now and a really fruitful collaboration, and that's all anchored around that protocol. nice. so,. You guys mentioned.

A

The .net standard client that you recently built.

A

so now, and we know there was old client,, but it is new client, because .net standard means that it should be able to run in .net core,.

A

It should run in .net full framework and a lot of other flavors and versions of .net,, so essentially now,. We could leverage everywhere that .net runs., correct. right? and we'll be able to work with service bus.

C

Yes. so first, we have a new performing sample that we just published.

C

okay.. That is now where I have just wrote the instructions, although I'm mentioning it,- and I mentioned in the instructions, for this- is how you run this on linux,. This is how you run this on windows.

C

right., because we're now liberated from the tie.

C

Into windows and we can go and make it work everywhere- whether .net runs on runs.

C

and so,. Christian can actually show you some, talk about the standard client a little bit and show you some. that'd be great.

B

I had two little samples.

B

one is,. We want to show it's end,, but let's start with the first one.

B

so,. If you know service bus- and you know the different clients,, then you also know like sometimes, yes.

B

We just made, if you compares them,, you will note that we made a lot of effort to make it much easier to interact with the new client..

B

So,. Here are little samples. so, you can see. Basically, this is how easy it is to register receive palm..

B

So basically, you say, "okay, register a message. Receiver.".

B

And at the same time, when you do that,, you also say whether you want to receive your exception,, so you have an exception: callback.

B

so,, you have a lot of freedom of how you treat and act on your exceptions.

B

So,, this sample is really very blunt and simple..

B

It does nothing but sending and receiving messages from a queue,, and you can see like you, have two methods: here.

B

One is processing the messages, one would act in case if would get any exception.

B

And then, the last one here is really nothing else, but sending like a few messages,.

B

So, let's just run it. okay. you will see, sends and receives two messages,.

B

But you can see, it really doesn't need a whole lot of code,. If you remove the comments, probably like some 30 or 40 lines of code, right?

B

so,, we have the internal problem.

C

Of demoing tcp/ip not being very sexy,.

C

Because you only send and receive messages., what's noteworthy about here with the choice that we made here, is that we actually took away, so going to .net center,.

C

We made a few api changes towards it compared to the old client, because the old client had all the features of.

C

All the services that jammed into one place- and we want to go and break that apart. yes.. Then it had significant dependencies on api. It's like.

C

Dubcf and other things that we completely took out and we re-plotted that completely on amqp.

C

And we use that change.

C

As an opportunity to address a lot of pain that customers had or that.

C

We had in terms of supporting those customers depending on how you look at it,, because some things were just too hard. so, for us. One thing we do know- and this is what reflects in that sample- is we now no longer have any explicit way to go and receive imperatively to receive a message on the normal queue clients.

C

on the normal queue client, you say,, you register a handler, and then you get called back and you process that message and when you complete that message, you are going to get the next message delivered to you., so we're dealing with all the complexities of a message loop for you.

C

sure.. If you want that control, you can pop out and you can use the message receiver., but we just try to keep everybody on the simplest path. So.

C

That we're dealing with all the possible error traps for you.

B

Got you., let me show you two things. one,, two things.

B

Mention one thing, show you one thing.. If you want to go on a lower level,, so you don't want that extra comfort. You just spoke to,.

B

You would need to use like servicebus.core here., you see that,. You would add that by using,.

B

And then you could go like for example,.

B

Instead of using a palm,, you could go to call receive directly and sent directly.

B

So, these things just like just a little trick. If you want to use this library.

B

And the other thing is right know the old process in new library,. The new library doesn't yet have management operations. Included., we all know gateway, include those,.

B

They will be there,, I would say by june timeframe.

A

So, a couple of months we'll have that ready to go.

C

Yes. yes.- that was a long, rounded thread on github. That was an issue filed and-.

C

Is that issue 65?.

B

Issues,, the infamous issues.

A

Is that the infamous issue 65?.

C

Yes. and so the thing is, we have this,.

C

There is this principle in azure that all the management operation has happened through the azure resource management interface., since our service is old,.

C

Our management interface far predates any of that.

C

So,, we always had kind of in management, in the service.

C

We took something away, which is very terrible. Always.

C

And we said, well, there's a way, so you could do this over here, and then customers came with, somewhat expectedly, with lots of use cases. But that said, it's not a fit. got it. so,. We had, the namespacemanager., however, that we're replacing was a little bit unwieldy and large.

C

so, we're making this a much more focused thing..

C

We say we really make a razor sharp focus on.

C

The exact users cases that people need to have the dynamism for and support those.

A

Got it. so,, hopefully, when that happens, it will be a lot easier for people that are using,. I guess, the older client that is used to that type of behavior..

A

Now it will be easier for them to come over to the new client. correct. correct. and how, like the same type of behavior from using that nuget package. correct.

C

Yes,, that's right., but in general, from a philosophy, perspective,.

C

We believe that your topology, that you're using in.

C

Service bus should be something that you'd consider part of employment.

C

that you would then use these dynamic entities. Sparingly.

C

Okay. and we're trying to reflect that philosophy in the api. right. got you. so,. One thing I wanted to ask you,.

A

So I generally I'm a asp.net developer..

A

I develop a lot of web applications. so, when you think about the request, right,.

C

Like the last time of my request is usually very short,,.

C

Works like a desktop client. right.

A

So,, when we're working with the service base client, is it recommended that we use it like as in a singleton type product and try to keep it long lived or there is something that is very transit and do I recreate one every time?, what do you,? What's the best advice, there?.

C

So, we do some magic under the covers to make it not matter all that much.

C

okay., because we use collection, puling., okay. great.. But generally you should ink a piece connection, oriented protocol.

C

And you're creating a, when you have a queue clients,, you build a relationship with the server.

C

Which means the first operation you do something, you create the connection.. The client then holds on to that connection. so,. You should go and treat that as a resource in your application and protect it..

C

It is threat safe,.

C

So it's not a problem. If you have a queue, client and then use that for multiple threads,.

C

Even at the same time, because it'll be fine.

C

so, we're dealing with that for you. okay. so,, you should think of that as a singleton effectively that you use once,.

C

And one's means you should be a little clever about it if.

C

You're building a giant one sort of it does several thousand requests per second, and then you should be a little smarter about it. Probably.

C

Do a client on a procession basis, clients,.

C

You know what a reasonable scope is. sure I got. You.

A

So, another thing I want to talk about a little bit about service, bus. so,. I know there's almost two flavors of service bus., there's a standard version,, there's also a premium version. right. so, me as a developer,.

A

Could you tell me a little bit about the differences between what I get from standard and premium? like? What's the value add of me, moving over to a premium model? so,, the premium service bus.

B

Is really for production.

B

Workloads in the sense where you, this is where we add all these features like for example,.

B

Recently, we just added the metadata [inaudible] we're adding data there very soon.

B

Right., these are things we really resolve for our premium deals.. It will not find its way to stand that,. So if you as a big insurance company or a big retailer or a car manufacturer- and you rely on.

B

Service bus and you need.

B

To do data recovery, write or geo metadata recovery, then.

B

These things will never be in standard.. You can build these things, semi on your own, right.

B

That's work and it's effort and.

A

I'll have to maintain that myself. right.

B

A few other key differences, for example, things like prefiltering, or that we're just implementing right now or as later be a planning a feature called audit. Trail.

B

All of these things will be on the premium tier, right.

B

so,. All of the things that you really need in enterprise messaging will never find its way to standard.

A

Standard, I think I would see.

B

It as like an entry levels queue.

B

you can use it to code on,. You can use it to test everything for very low volumes scenarios in production. You could use it.

B

whenever it is about production and you want a high availability and you want guarantees for throughput and you want repeatable performance,.

B

Then you should look into premium.

A

Sure. so, like you're saying,, so if I get premium, I get that guarantee in terms of disaster recovery., I'm sure I'll, probably get more more throughput.

B

Availability zones will be therefore,.

A

We're implementing that right, now.

B

So,, all of these things will be on premium,.

B

We're not putting interest on that. got you. so,. If I was suppose to use standard.

A

Would that be good for me to do something like,, I had a simple task here or something like that.

B

Could I use the standard version for that?.

A

Yes., theoretically, you.

B

Could do that and the session features of another factor helps a lot for that because see a session like a sub queue right?

B

yes.. So you can use that to split your workload for the regarding tasks.

B

I would think storage queue might be better for your task. Queue. Scenario.

B

would you? okay., but you could use it,, that's possible.

B

so,. Why would you say that?, so why would you think so?.

A

What is it about? The features of storage queues versus.

A

Service bus that might make you feel like that might be a better solution for that?

A

storage queue. You can have easily more queues.

B

Okay. and is very lightweight, right..

B

If you look, for example, at the storage queue,, the message doesn't have headers.

B

it's a pretty simple format, right.

B

And task queue is typical of what you need.. It's very simple, and you just need like,, hey, I'm going to pack my task till the other side able to pick it up.

A

So,, I don't need some of those features like filtering,.

C

Disaster recovery, location things of that nature.

C

And you very often have the duality of,.

C

This is why it's kind of associated with storage.

C

is that you are throwing some job.

C

Represented by some file into a blob store, and then you say,.

C

Please go and process that someone. so,. It's logical that like, if you already have a storage, account, you're in the storage, account that you're storing that file and then you're putting a task into the task queue that goes and processes it. Up.

C

so,, makes sense for a lot of those scenarios where you have something.

C

That's literally just a simple task. You want to go in and fan out to a bunch of workers.

C

for that it's ideal. and then, once you get more sophisticated scenarios, then.

C

You're landing automatically more or less service bus, territory., right. so, like when I want transactions.

A

And rollback and things of that nature,.

C

That's when that will be coming? and scheduling,, there's thousand features that we have and then you have in queues support, obviously for interoperability. so,, there's a lot of reasons even to use standard.

C

What you get with premium is really predictability, because the resources associated with.

C

Premium namespaces are yours.

C

they're reserved for your reference, all the time and in standard you can get a shared pool and if lots of people are busy on the same cluster, where you live,.

C

Then you're getting less time. got you. so,. Did you want to.

A

Talk to us about some of these features that you have up here?.

B

Very briefly: walk through it., so on service bus.

B

One of the features is scheduled: deliveries. so, you determine when then the message should be actually enqueued. dead, lettering.

B

so,. After a number of retries, or after a number of failed attempts,, the message gets dead lettered. we never lose a message,. We just move them to that other queue for you to inspect it later.

B

forwarding,. So you can determine between,.

B

You can say from this subscription to this queue I want all to forward., then deferral..

B

This is like a workflow oriented feature where we could say, oh,. This message came to me out of order..

B

I want to put it back and the I will just process it. Later.

B

sessions,. Imagine it like, as queues and queues, right.

B

Like little sub queues, very like weight, you can basically use that to enforce first in first out, order., batching and long polling, meaning this is,.

B

We talked before as clemens mentioned,. Oh even grid, for example, is allowing you to do pushing instead of polling.

B

So, treatment, therefore, for service bus are to integrate with event grid.

B

And the second one this we allow you to do long, polling,.

B

So you have my connection to the broker..

B

If there's no message within 60 seconds.

B

right?, and only then we do another request. yes. so,, we request only every 60 seconds, with classic long polling, there. batching,. You can send multiple message and longer transactions.

B

so, basically,, it's like if anything, fails,.

B

Roll back, filtering and action.

B

so,, you can basically apply filters and that's actually the last sample I have for us today.

B

they say, oh,, you processed a message in some certain criteria and now I apply filter and send it somewhere based on that filter and then auto-delete on idle. so,. That's the subscription which never would get messages.. You could delete it or have it auto, delete,, a little dangerous, feature., duplicate detection. That's based on the message. Header.

B

So,, if you get two times the same message either we will detect that within a certain time, span.

B

A

Then role-based authentication.

B

And msi support that's in previous stage right now., but it's going to shoot very soon. the geo-recovery we talked about. develop across platforms,. We talked elaboratively all about.

B

so,. We are a very open source, supportive.

B

And then on the right-hand side,, a few of the integrations,, I say it by now is azure event. Grid.. We have very tight integration with logic, apps functions and dynamics, 365 for sales,.

B

Actually, a pretty solid integration story with us.

A

Cool., that's sounds pretty good.. How will we just take a look at the demo that you have. yes. and take a look at some of these features. so,? That is the demo which shows you,.

B

Oh, I get like a message from some web shop.

B

On the left-hand side,, it just basically sends a message.

B

It comes more like a vibrant,, no headers message.

B

So, you need to process as metasearch,.

B

The one has a way and figure out, okay, what's going on in there?, and then you basically reroute it for as a processing to a subscription,.

B

That's the second part here. and then,, on the way of doing that, we also apply some filters.

B

yes.. Let's take a look at the code.

A

So,, that's basically,.

B

The sender codes we have used to run it., we can start the lecture., and now it's just like. I made it into some earlier functions which just keep sending messages just for demo, purposes.

A

Okay, so hold on so one second. so, your azure functions has a service both clans on it.

A

And it's going to be sending to a queue.

B

Exactly. okay. exactly. so,, we have two functions. one is just sending and one is receiving and then sending to a topic. Basically.

B

so,. I show you both.. I show you a queue and I show you a topic with subscription.

A

Okay. so,, I don't think we've talked about it. but so, really quickly,. Can you tell me what's the difference between a topic and a queue inside of service bus?.

B

So, a queue, is really like. I think the industry term is a message: channel. okay., where you would receive and send messages one inbound,.

B

One outbound, and then a topic like pops, up.

B

So,, I can have multiple folks.

A

Listening to the same topics., you have a topic and then end.

B

Subscriptions for the topic. and then you can have each subscription gets a message,.

B

Basically, a copy of that message.

B

and then you can have filters on the way between topic and subscription and rules and actions.

B

so,. You can create a rule which has like a filter. so,. If that applies to that action and then basically move it to that subscription. cool. sounds good. so,. Let me just start it up.

B

So, we have something to look at.

B

This is basically how everything looks like under the road.

B

so,. Here is my queue.: that's where this stuff will be coming inbound., and then we have basically my topic here and two subscriptions.

B

Yes. basically high-value and regular sales here.. I called it like that.

B

on my code is doing really,.

B

It sends first who's. This queue.

B

It sends a message containing only two fields.

B

Obviously it's just a simple sample and that's one time: json,.

B

Sorry,, that's just a customer and then the sales value,.

B

Some random sales value. and then basically here, that's an interesting thing because clemens, he mentioned old, client and new client, quite a bit.

B

so,. If you want to interact between multiple service bus clients,, you typically want to send a memory stream. You don't want to just directly serialize into json, and then abide.

B

But you want to use a memory stream, so this code is kind of resemble how you would do that.

B

got it. yes. and then on the other side, we receive that message, basically,.

B

We receive it in batch, so we mentioned the feature batching. and then we basically,, instead of keeping to send just the body,.

B

We add some header, we say, oh,, let's add to header. sales value, here.

B

So, my system, which is in late, it's end processing. The message for us can react to that sales. Value., let's see, if that all works., we got a few messages. Here.

B

We've got a few messages: here. so, the function basically started pumping going in here.

B

And then you can see. We have two filters here which say, oh, I have value sales,. I want to treat sales which are more than $10,000.. I want to treat some difference and regular sales.

B

so,. You have the opposite: filter, basically.,.

B

Yes. and that's something.

C

That's been processed by service, bus. so,, you send the message. In, message has that header.

C

Your service bus, never look at that message: body.

C

so,, that's something that is effectively protected when we never have a look at it., yes.

A

So,, what christian is doing, here,.

C

He's taking a messages and he's promoting effectively some information that sent out into a message property, and we can then go and take a look at the message: property for routing purposes, as it does here, and got it.

B

The cool thing is, you, don't have to write any code for the actual promoting of that., that's right.. That is also response. So, yes.

A

That sounds pretty good. so,. If we wanted to take a look at the sample,. Is this on github somewhere?? Can I look at this or are there other samples that we could take a look at??

A

This particular sample,, not yet.,.

B

Okay. it's fairly new,. I just worked it.. Is it private?? I will actually put it on github and it will be exactly on the same place where I show you right. Now.

B

So,, this is our standard github label,, its azure/azure service bus., and then this would get you to dev client,.

B

That will get you to java client., here's our sample,, all the group in .net java and management. so,, all kinds of management operations or sample for ad management operations. You can find here.

B

Samples how you could consume azure, metrics,, the monitoring, metrics pipeline. and then here,. You have all kind of samples for the old clients, for the new clients,, something very more to primitive samples to get started.

B

And then the puff sample clemens has mentioned before,.

B

I has link to it: here. so,, you have that available, too, and then interesting as a clemens's mentioned, or you actually brought up to christian.. We would support jms or rather you when you speak, amqp right?

B

yes., then you can induct the service bus. so,. We have like jms sample here and we have more.

B

Like service bus samples in here for java.

C

And we've got samples coming for.

C

so,. We have brand new clients, coming, brand new implementations that we build of amqp.

C

Sure. and then putting clients on top of it for a number of additional languages.

B

So,, that's all in flights.

C

And we're making those things better., you can use, go python, et cetera already today with the proton climb., but we are creating officially bless and microsoft supported, api.

C

sure. wrappers around thoughts.

A

Yes. and sometimes completely new amqp implementations.

C

Great. so,, that if I wanted to.

A

Get started with service bus in .net or in java or other languages,.

A

I can effectively come to this ripple.. Take a look at the sample.

B

Yes., I'm sure we also have really good docs, that.

A

Docs.Microsoft.Com that we can take a look at too.

B

We do. so,, if you go to the docs.microsoft.com and in azure service bus, messaging,,.

A

You have basically overview.

B

Where this service bus, some fundamentals, and then a few quick starts, where you can say, well,, this is how you use queues,. This is how you use topics and subscriptions.

B

and then from there,. These links here like if you follow these samples,, you see azure. We have it for multiple languages.

B

So,, the news will be go then: a new node library, a new python library.

B

Sure. and then, yes from here,, you will also get to github.

B

Great. well, this is great.. This is a good start.

A

I kind of feel like we have so much more. We need to talk about. so, we going to have you guys, come back on again. Another time.

B

Another time. yes. that will be wonderful.. Thank you., that's great.! So,! Here we have it.

A

That was another episode. On.Net show where we learned all about using azure service bus.