Istio IstioCon 2021, 10 Mar 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Istio Cookbook: Kiali Recipe

Description

#IstioCon2021

Presented at IstioCon 2021 by Lucas Ponce.

Kiali is a management console for Istio. It provides dashboards, observability, configuration and validation capabilities. This workshop will walk you through practical examples of Istio using Kiali.

A

uh Let me start with a thank you for with the with the last slide, to give you a a a thank you for for joining, for, using for using this technology still for using kiari to see this technology and also on top of what pedro say. You know, if uh all what we are going to do with this, um this workshop is public. It's well documented.

A

So if you cannot finish today, also because it's late for for your time zone or or do you want to continue their day, we are also in the in the ali room in the istio slack and as all developers from all generation. We use ifc as well, so you can find us in the easter slack hilly roon or if you use irc, we also love ioc and- and we are in the room on the on the free note to that right, so yeah, thank you all for for attending.

A

So, okay, let's start what what we are going to do, what you would expect with with this demo. So what we are going to do is to install a kubernetes and istio cluster. We are going to deploy demo applications beyond the hello world beyond the book info right. We want to do something a little bit more complex that can simulate more um what could be a a more realistic situation, a more realistic scenario that you may have, or or the reason that you are investigating or using history right.

A

You have some real applications, real problems, and you want to see okay, what are the necessities? How I can start using this technology right? How we will do? I will be using the google cloud platform to create um a cluster for me because it's pretty straightforward, I'm a developer. It's quite straightforward with your gmail account just to enter in the console, create a cluster, but the documentation is completely agnostic of the platform.

A

I think that the documentation is very well uh documented how you can do the very same with mini cube, but other platform is perfectly fine. You know to to you, so there were some resources. What would we use? So we are going to use the the the east iowa getting started, so we are going to say. Okay, you are landing into the technology.

A

I guess that if you are here at least you you have performed this step before right, we are going to change some steps from the getting started. For example, we are not going to use the histocity last four we are going to publish uh or suppose the kiali service through the ingress gateway. We are going to show how to do that right. How to do that inefficient.

A

um All the use cases that we are going to show, and also in the kiyali dotaio tutorial, so you enter into the tutorial and you will see all what we are going to do is: oh sorry, okay, so feel free to navigate just the slides to click into the urls right. um This is basically all that we are going to do step by step, explaining this stopping on on each steps, trying to answer your questions right and, unless you to understand basically the easter technology in a visual manner.

A

This is what is the goal of this of this tutorial okay platform choice, so I'm going to follow up all the stuff with the google uh cloud, solid, the the google cloud platform. Why? Because I'm a developer, I have a gmail account and to me this is the easiest way. So I have a project. I guess that you also have a project. um It's quite easy to get a project in google. Is this the first time? I think that sorry, a disclaimer?

A

I I'm I'm not expert in this platform, but at least as a developer, if you're the first time that you sit, probably you will have some kind of free budget for producing, which could be pretty much enough to run all the tutorials. If someone would need um an additional project to share a project, just let us know offline and we can try to enroll somewhere else into one of the demo projects to do that right, but this is what we are going to do.

A

Okay, so um let me see okay here, so uh this is the only specific step that that it depends on the platform. I don't know which box do you have so um in one of the steps we want to create, as I said at the beginning, uh a more complex application beyond the book info right, so we want to use real or something that is close to our real dnas, because we want to set a different host because you know the gateways can be changed discriminated by the the horse resolution right.

A

We want to simulate that. So uh when we design this this version, we have two options. One is to use real dns like we can do in the cloud platform, but that's not a problem, but it takes to to refresh the cast and refresh the google servers, the dns service. It takes something of 15 minutes, 20 minutes, and I and I thought that that could be- uh could make the works a little bit more boring waiting to have a dns resolution.

A

So instead to use um a published dns, we are going to use local dns, updating the etc right, and the att host is a mechanism that is valid for linux operating system, macos, right or windows right. So if you are not familiar with these steps, just let us note in the in the in the chat, and we will tell you just how to how to do that right. This is the only requirement that you may need right also to change that from the very same place that you're going to open your browser right.

A

So if in case that you have some bastion machine which you are using to do something remember that this is to resolve from your browser right, a specific ui. This is what we are going to do. Okay. So what is the agenda? What is the purpose to do in in this two and a half hours right?

A

So, first, we are going to prepare a kubernetes and an easter cluster, pretty much following the getting started, but with some additional changes like exposing the addons from the gateway instead to use the cube, ctl or forwarding, which is the the standard way used by by ctrl command, we are going to to deploy a more realistic application than the book info, which is called the lens traverse demo.

A

We are going to deploy without mess just without like in a real scenario. I have an existing application. I deployed to my cluster right and the next thing that we are going to do is to to to describe what could be the first steps right to incrementally, adding and rolling my workloads into into into the ms and see what are the features, the capabilities that istio is providing me right and how I can see these features into chiari.

A

The next thing that we are going to do is to learning what kiali can do for you. What information istio is collecting in a low level, metrics tracing logs configuration and how chiali can abstract that way. I can show you all this information all together in a way which is easier to you to understand what the cluster is doing. What the service message is is behaving so what potential problems may have right, and also we want also to do.

A

We have time to to give you some uh some hints on very powerful feature that keali can give you, for example, to query your cluster to see hey, which is the slower pass, which is something where should I put my focus when I have a highly populated uh set of warlords to do that?

A

The next thing that we are going to um to do is to to practice the the histo features right, but using kiari right. You are going to send me. I guess that I I think that you are all there. If you are watching all the easter con.

A

I you can tell me that you have seen a lot of jammer files, so you will see me that in this workshop, probably will be the the only conference or session where, with the less damage gamma file text that you are going to read, I can I can tell you we are going to do all the steps of both of the stuff we are going to do through kiali in a visual way right how to shoot, but um last, if we have time so we also try to to to to progress into the more security, more advanced scenario.

A

To do that. As commented, all the the steps are in in the tutorial you are in. So it's something that that you can do in your own reason. You can try to to follow by your own. You can watch me or just to do to attend the explanation that we are going to do that, I'm going to start from the scratch right. So I I guess that you can enjoy so any question.

A

Let us know in the q, a in the in the chat or how pedro has has commented okay, so uh the first thing that we are going to do is the prepare, kubernetes and istio cluster. So we are going to use the insta.io platform setup and then getting started uh as I don't want to teach you how to browse this documentation. I prepare a workshop chipset in this url.

A

So if you open it, it's basically the very same thing, but in a way which is more easier and faster just to create a cluster right like open uh the console cloudgoogle.com with the gmail account. Okay, I have that the next step. What we are going to do is to create a project. If I have not a new project, so it's something that I can do here a new project.

A

I have started this for me. I actually you have a system project, that's okay, and what we are going to use is activate the cloud share, because in this way we are going to skip all the the installation, steps or the troubleshooting, for if you don't have install qctl command or if you have a different version of hem or etc. So this closure is quite powerful in order to when you are running this kind of workshop, because you avoid all the travel shooting. If the path is correct and.

A

Etc. Okay, I have it. I was crossing fingers because I told pedro the only thing that I cannot ensure is that google is down so, but this thing that that now is responding right. So I have my share right. uh Yeah. What we are going to do is to prepare a kubernetes cluster using the specific platform setup for google cloud engine, describe it in istio, so basically I'm taking I'm just copying the the same as to save you some time when browsing that so I can get this export and I can copy on my.

A

On my share on my cloud, shell, and what is going to do is populate a cluster right, I'm just taking the defaults.

A

On okay, so in this step it's quite basic as you as you can see, we are going to check that the the the cluster is correct in case that you have a linux box and you want to do that with mini cube in the step one to be from the from the cheat sheet you can also to to to get the very same thing, but for a mini q box.

A

A

My cluster is being checked. I guess it's two days it's faster than yesterday. Yesterday, when I was doing some some demo, it was running quite slow, but today it seemed that this is okay.

A

If you have already a cluster created, you can use it as well. So I think that the what we are going to do is to generate new applications, so that is fine right, but I think that what we are going to do for from our site is to start from the scratch. Just to show you all the steps in case that someone has any problem, we can help him help her in a faster way.

A

Okay, master is healthy, that's good.

A

Probably for people living in the in the u.s zone, uh this could be faster than me. You know I'm from europe and from the euro. Folks, probably this could be a little bit more slower, but okay, I have it so this is my cluster.

A

I I guess that's more or less you can be on the same on the same step. So uh what I'm going to do is to take the details. Okay, it was generated and I'm going to do a very basic cluster test to asking.

A

The notes of my my cluster, so a standard configuration for this cluster following the steps from the istio platform setup, is to create a four nodes cluster, which is more than enough for ornith, okay, so uh yeah. This is a step one dot. A of this section. So in case that you are using mini q, you can follow the 1.1.b, uh so what I'm going to follow now is to install istio. So in this case it's pretty standard. I know that there are a lot of um options to install still more optimization.

A

We are going to to to use the the the the most basic way to install it right, probably in a future session, if you like, we can try to to create another session for more advanced scenarios. Installation scenario of history with kiari.

A

So uh again this this cheat is just to to save you to navigate from the from the history site. What we need to do is to download the istio with this command right. We are going to use the istio das 1.9.0, so we are going to enter in the in the istio folder.

A

This is important um yeah for people that are too fast, because the clouds, the cloud shell by default is shipping, history cpl, but an old one. Probably so, if you are, if you don't update your path and check the htcpl version, you may have an istio, 1.8 and yeah, and that probably you may have some same behavior with a if you want to use htctl from an older version with another historicity, with a newer version in the project in the cluster right.

A

This is the only step, but again this is what is documented in the in the in the in the base documentation. I say just to update the the path to to check that you are using the one dot my version right and to install the demo profile. Is this in ctl install set profile demo right? This is what we are going to run to start installing the cluster.

A

If some step, I speak too fast because it's less for me, and probably I I I run so.

B

A

Let me know from the chat: hey you guys. Could you repeat, or could you stop? um That step was not fully clear? Yes with you know, we are doing a workshop. It should be uh typically we like to do that in on-site events, that this timing to be virtual, it's good because we meet with a lot of new new folks but yeah in case of interactions. Oh yes, just let me know guys we'll be happy to to clarify something: okay, so, after installing the demo profile the next step, this is typical.

A

We are going to install the demo app. The booking folder has to check that istio is working correctly and we don't have anything extreme from the platform right. So in a nutshell, these are the two comments that we want to do is to label the default name space with the istio injection enable in one comment I'm going to clear my screen and in the next comment, just to apply the typical booking form.

A

I'm quite sure that if this is not the first time that you install istio, you have done this several times before right, it's nothing, new, okay, uh yeah and then uh the next step that we do is to check the the pots with uh dash w to check. You know to wait until all the pots are are in running status.

A

Okay, reviews versus three his reviews is now up.

B

A

We have two reviews: one details, I'm still missing the product page and also the ratings.

A

Okay, we have the ratings and we have the space okay, so the booking info should be up now so uh to test the demo check.

A

We are going to run this command from inside a boxer so just to to have a first check that that the pods are are okay and yeah. It seems that the booking phone was successfully installed okay, so this is something that we are going to do a little bit different. So in the this in this chipped sheet, we have a three steps.

A

So um in the steps three, we are going to open the demo up to also traffic. So uh this this one just take a look of the step, because is when we are going to change the way that we typically access to book info and we are going to publish the book info also from uh from the right we with a host right. So to do that we are going to install the gateway.

A

Remember, the gateway in istio is the way to tell okay. I want to expose a specific service through the gateway. So now this service is is ready to get traffic from outside of the cluster, but yeah from outside. It doesn't mean okay, how I can access that so and then to do that we need to uh in in a real scenario, I would ask to to to my network admin to tell a for my new application. I knew a new dns. I knew a new name for my new application.

A

Then this will be there, the name that I'm going to access to this host. This is what we are going to simulate now right, so uh in what we are going to need is to uh to get the ip of your ingress.

A

So that we can check- which is the external ip of your english, take a note that this may require some platform specific, depending on the platform that you are working. If you are in google cloud platform, it's just uh the external ap of the easter english gateway and just to show you how I would update this.

A

uh What are the next steps for my application in a real scenario in a real scenario, with this ip, I tell to my admin guy or my network admin guy, so hey, I need a new dns, and that means a guy so go to the to the networking services.

A

Go to the cloud dns and probably he will access to us all and we'll try to create a new entry with your ip. We are not going to to do this for the warsaw, because, okay, this is it's quite simple. You know you need to to have a public domain.

A

I need just to create a a entry for their scenario, but instead to do in a public mode because it takes you know: 15 20, minutes to refresh all the servers, but we are going to simulate that from your browser, so um I'm in the cloud shell, but I'm going to open a cell from my my linux box. Okay, this is my machine where my my browser is running and what I'm going to do is to uh to edit the the etc right.

A

This is the way that I'm going to simulate that I can resolve a host from my from my from my browser right, um okay, so what we are going to do, we are going to create a new entry with the ip of my ingress to the book info. In this case, we are going to use for the tutorial the dns that we want to do right. We are going to put my ip, my ip is this one?

A

Okay, let me show that looking for uh your blaster.

A

Okay- and you should have something like this right for your case- you need today with your cluster english ip, and this is mine right um once we save the this file, uh we can check that we can resolve the booking for from my browser.

A

Okay, I have the working voice- it's not complex, but they can understand that and depending on the of the platform that you are running windows linux mac, this could be different okay, but this is looking for and I'm accessing through the ingress, but I'm not using the ip I'm resolving the host right. This is what we are going to do: okay,.

A

uh Okay as well, so we, I think that we have is to install. We have the demo app. We check that the demo app is working. We expose the booking force to the gateway. We link the ingredient, the booking for uh dns, that we can resolve from our box, and we explain how to do that in a public scenario.

A

Okay, the next step that we are going to do is start installing the diamonds right we are going to follow the getting started, documentation to install the add-on, just simply to copy paste and just to deploy the samples add-ons files that are shipped with.

A

B

A

So you can check the pots of the system namespace.

A

Okay, what's fast, you see that everything is is running, which is good right. Okay, now this is the the this step is different from the getting started here we are going to do something documented, but different.

A

So probably you will see that in in the easter there is a section which is in the observability test, which means we are going to expose the addons in specialty ali through the english right, and we have two options to do that. We can do that in a very secure way or an insecure way right and one of the requirements that is asking you is to to have a previous defined english domain. The english domain is like the istio cluster or that that they choose to do that.

A

So we are not going to create a public domain, but we are going to continue using the htc host mechanisms to simulate that I have approached the main, but I want to expose really through them to the gateway. So what I should do so again I go to my epc host. I will repeat how I get here. So in my case it should have been easy. Hospice you know in the slides you have described how to do that in max or windows.

A

Okay, um the same ip, because it's the same english service that I'm going to reuse, I'm going to copy this and I'm going to create another entry with the same domain. The domain with that I'm using locally is called steeldashcluster.org.

A

But for keali.

A

The very same thing yeah, you can see it.

A

Okay same ip, because I'm using the same english ip, but for tr. So in this case, uh what I I would like to do is to expose clearly not using this just ctl dashboard, because remember I'm using the cloud shell, which is in a bastion machine of the google platform cloud, but I want to pause it through the ingress to to do opening from my browser right. This is the step that I'm doing right.

A

Okay, now that I have this entry, what I need to do in my in my cloud, shell, is to export this english domain.

A

A

Okay, I'm using the istio.cluster.org and remember and the the the very same thing that I did for the book info was to create a gateway for the booking for and a virtual service, basically to expose this service to the ingress and also to redirect all this traffic that is coming from this host from the english to this particular virtual service, also to to connect to one right, so we are going to the very same thing and yes, this is not not new. This is also well documented in the istio site.

A

You know how to expose them, how to expose them the add-ons through through the english, but in this in this, uh which is, is going to save you some some time right.

A

Okay, I copy this.

A

Okay, I generated the gateway, the virtual service and the destination rule that clearly didn't have. So I get this easter obvious and now.

A

I should verify that from my browser I can access to kiali from.

A

Kiali.Istio.Cluster.Org.

A

Okay, so if you are here, congratulations because you have installed a cluster, you have installed istio on top, we have installed the booking for demo and has installed the add-ons, and now you have exposed kyali through a public dns. Okay, that's good.

A

uh Most people are on on this step, just to do a a quick sink and let me do to drink some water.

A

Please, okay, I guess no comments on the chat, so I guess that we are good okay. So, let's.

A

A

So uh we are going to start doing a key ali hello wall right. We are going to start using kiali in you know in a in in in an efficient way to detect something that is at this same to our scenario, so uh we open key alley and when we open key alley, we learn into the overview page. The overview page is telling me how many name space I have enrolled into my mess.

A

You know how many namespaces are in my cluster if this name space, what are the labels that they have right and also I can see a first thing. Do you see what I am what I went to, but I see that they have a couple of warnings here, because the key ali has detected a you have here, a default name space- and I have here a couple of ones and also you have the name space and you have.

A

I analyzed this geographic object, but I have one warning: I'm going to start with the default one. If I click in the warning icon, I'm going to move to the easter config page, it's going to list all the easter configuration that I have for this name space. I can select one too many right. I can select multiple, namespace and and also filter.

A

uh What could be the configuration with warnings for examples and I'm going to uh to check the booking for gateway that it has a warning.

A

And what is telling me that? Okay, I have one in here in the house, and I have a wild card. Ankial is telling me hey. You have more than one go away, sorry more than one gateway for the same host sports combination, and if I clicking the show overview you're going to tell me. Okay, you have the book info using the gateway and the key alligator was the same thing. If I click in the key allocate way, I see that okay, the key alley gateway, has a host, but the booking for gateway.

A

I has a wildcard what means when I'm sharing the same ip?

A

Okay, when I share the same ip, I need to tell the istio that hey, you need to discriminate, which request goes to chiari and which records go to the to the book. This is the war of the english right, so one of the of the of the basic ways to do that is to define in the host on this. In this case, you see that the booking for gateway, we change the the book info to use the booking for easter class centers.

A

So this is the host that we need to update here into the engage we are looking for. We can do that from kiali. I should print here looking for your cluster.

A

And this is the only xml file that we are going to use on this workshop if you expect it to deal with a lot of files on on these two hours.

A

No, this is no way we are going to try to do everything in a visual way now in a more friendly way right, so I saved the gateway from kiyali and now, if I return to the review, I see how the is uh is now in green, saying that okay, it seemed that everything is looking okay right. So if I click into the easter config, I can see all the uh the history configuration that I have defined into the default and the system.

A

You know the extra system typically has more specific default foreign filters, but typically this is what I would see in my application. So the virtual services, the gateways or security objects, are affected to these workers.

A

Okay. So if you are on this stage so uh hello world to kiali, welcome to these applications- and I think that we have completed the first, the first step.

A

Okay, so uh we are going to start the second point. Yes, let me let me know if, if, if I'm fast or I'm dying, so use the chat to tell me any comment and you can try to juice.

B

A

What we are going to do now is to deploy um a new demo app. These demo applications uh consists in three naming space and a set of microservices deployed on each base office. We are going to deploy them just to to copy and paste these instructions from them from the kia tutorial page and go to your cloud. Shell.

A

Okay, I created the name space.

A

We are not going to to label anything, so we are not. We are deploying the application, but we are not, including or on hold or just saying, english if you are including or rolling or connecting to the mess yet so we want to. You know to simulate a real scenario that you you have some set of services in your kubernetes cluster, and you also have installation and you want to hey- I want you, you want to learn how all this is your feature, works and and to see what they are doing.

A

So uh we deploy these these applications. um We can see how the pots of travel in c name space are.

A

Oh, quite fast, are working, but let me to spend some minutes just to to explain you the design of this application, what how this application was saying. So uh so uh we have um a travel demo that simulates two business domains that are organizes in basically in two name space. So from one name space. What we call travel portal, we are going to deploy what we call several portal shops.

A

Those are intended to be ui shops right when the user can search. I can can book flights, hotels, cars, insulation, whatever they need, so um one of the features that we wanted to to design is that every shop can be hosed by different countries and may have different type of needs. So, at the end of the day, they may have different characteristics like okay.

A

Some countries are more focused on new users or assistant user on on john young customers or more other kind of senior type of customers or the channel to access could be just web or mobile, or you know, or they may be interested in one type of traffic versus other right. So the idea is that these shops can generate quite heterogeneous traffic, and we can you know to prove how east you can be in this situation right. So this is what the travel portal is is going to simulate.

A

These ui applications, one per country with its own characteristics. um This all ui services from the travel portal are going to talk with uh another namespace. What we call the travel agency with a basically a travel service and the travel service will be responsible to basically to tell okay what are the available destinations for europe right um this or all the the openings, and also to let a specific user to choose one destination, one traveler: they prepare a travel for paris, preparatory for berlin for lisbon, madrid, etc, etc.

A

Right, so this travel service is uh in the travel agency name. Space is going to be responsible to talk with one of many other back-end microservices, for example, you could take. You could talk with flights, it could play with other microservices with cars to insurance, which depends on the type of of the of the travel and also depending on the user. It may be accessible to a discounts and additional discount services. This is quite similar to some booking for logic.

A

Just to to to show you something more familiar and the travels service will be responsible to collect all these quotes and build an offer. You know offer quotation, that's to do this and back to the to the portal shop right, so this portal shop should be uis, but in for this case, are our simulators that are generating this random traffic to the travel science right.

A

um We have a third name space, which is called travel control that is going to to to work like a business ui right, a business ui that is going to to tell me how many shops I have is going to tell me what is happening. What this request to these destinations, how they are flowing to that in which destination are new, which portal shop is doing more requests, what the type of the users type of device type type, and also in the in the left in the left side of this ui is going to.

A

Let me to um to change the settings of this travelshop simulator. I can change the request ratio. I can one I can simulate that one shot is faster than others, and I can see I can also to to simulate that it generates one type of traffic to others. You know the idea of this is to to add a little bit more complexity uh and also to to try to add a little bit of more.

A

You know information or something more, but even if it's a demo app, but it can simulate more close or more realistic traffic that you may see in your scenario. Okay,.

A

Okay, so yeah this is we we have deployed application. I guess we should have all the the the wall laws employed in your cluster. So uh let's say to to go to the do the first steps I'm leaving still right, um I have an existing application, so what I want to do to do is okay. What next? What key ali can tell me on this scenario? So if I go to my kiali, probably you are going to see the same.

A

I can see that the the the the three new name space generated the travel agency- travel control reporter right. But if I say okay, that's that's cool. I would like to go to the graph okay in the graph. I I don't see anything. If I change the name space, I don't see portal.

A

I don't see control what is happening, why I don't see anything so the answer is that I have deployed my applications into the cluster, but my applications are not yet into the mess, so why I wanted to start this version with this scenario, because it's pretty common that when you are starting with istio, so you may have a similar situation that hey. I don't see what I expected. I don't see a graph.

A

I don't see my wallet in my we may happen so keali can tell you, for example, that in your applications you have the missing cycle. So that's a little intro that kiali is trying to collect all information from the for your world from the kubernetes api, the pods, the controllers from the istio telemetry, all the parameters, low level data collected from employees are collected and manipulated by the telemetry companies and also tracing information.

A

If you have enabled tracing on your on your set, I'm going to fetch all these tracing signals and I'm going to combine to you to show you this information. So this is pretty common, especially when I'm deploying a something didn't work. Well, so uh I'm I deployed application. My application was not connected to my cluster, so yeah this application.

A

I might I enter here. I have the workloads, I can click in a world. They can tell me hey this. um The warlord is missing the sidecar and I don't see here any history in its container. This is just a for example. A warning that hey this bot has no issue cycle is going to work, but it's not connected with the mess. It's not going to tell me additional information from that right.

A

Okay, so what are the the next step that that we want to do so or or what is something that typically we do is not to. One thing that is different from book info from this demo is that book info starts, adding all the application into the mess. So we want to do that in a interesting incremental way in an interactive way.

A

So now, for example, as I described, I have a first ui- and I would like to to put this uh this this to id service into the ingress just to expose it to the external traffic. We are going to do that, so how I can enroll a name space into the ms right, so I can go to the to the name space to the travel control and I can click enable I auto injection label. If I click in the label, I'm going to change my view.

A

I can see all my name space in a compact way in a detail view on a table view, and I can see that now, the travel control has. This injection enabled right.

A

So uh what this means, when I add the label on the namespace, that means that the new deployments, the new deployments, are going to be injected automatically in the sidecar by what happened with the system deployment, the system deployments are not going to change, so the next thing that we are going to do is to go to the workload. Do a system? Sorry uh only the tower control.

A

Okay, we are going to the control warlock, which is telling me that is missing the cycle and we are going to take okay. I want to enable auto injection from this workload. Again, if I redeploy the from qctl command automatically, history is going to detect that the name species level, and I'm going to do this by me. So as this is an old warlord, I need to enable manually this the the annotation to tell history that I want to put this warlock into the mesh right.

A

So what is going to do is that it's going to restart the report.

A

Now I will will show me that I have a new pot right and it's green with the the easter container. Okay.

A

Okay, now I have the the control warlord, the control service exposed to the uh sorry connected to the ms and now what I need to do is to expose to the demo traffic. So what I need when I want to expose an application, we need a dns entry, so this is something that that we are going to do so. The first thing that we need to do is to do our network admin until hey. I need a new entry, a new anc for my dns, for my control application.

A

This is something that we are going to simulate with or etc just to to save time, because it takes some minutes to and I'm going to get a new entry again, we are going to reduce the same ingress ip, because we are all the traffic is entered to my cluster by the istio english gateway. So this is why we are using the same public ip, but on this case we are going to change.

A

We have used, for example, a control, travel control.

A

Istio cluster or.

A

Okay, I have my then my dns entry, and, if you remember what we need to do now, we need to generate the gateway, the people service and the destination rule. If I want more security configuration, I will need to generate additional listio config files to to add this security thing, so how he ali can do that by me. So if remember that, when we are adding easter configuration, we are referring always to the services that we use right. So we go to the to the service to the control service.

A

In this case, anki ali offers to you a set of high levels uh with us to generate uh istio, high level extra service message scenario for you. So on this case we are going to to use the, for example, the the request routine on the control service and we are going to to create a basic rule, a routine rule. So, uh on this wizard we have a first builder on the tabs request. Matching I want to match any requests.

A

uh We can route to the one to all the traffic to the control workload forget for the moment the full injection request. Timeos, we will came back this later and I'm going to add the rule right, don't get it yet, because one thing that we want to do is: okay. Now, I'm creating this uh record routine rule, but I need a great way to to expose this routine to the to the outside traffic. So I can do that using the advanced options. I click and show advanced options.

A

I go to gateways, I can click at a gateway.

A

I can select assisting one or create a new one, I'm going to create a new one and now here in the gateway host, remember what we did with the book info with the kiali. I need to tell that this is the travel control. Is your cluster okay?

A

So if this is correct, if I didn't type on anything, so no mistakes- and I click on create so now keal- is telling me that is generating configuration for me from the service.

A

I can see the workloads that are linked with this uh in the selectors of this service. This is the the control workload and also you can see the hdr config that has generated a first virtual service, a destination rule. If I click in the http config, I can see that the ali has generated this virtual service destination, the destination rule for the control service and also the gateway okay, let's test it so go to a top of your browser.

A

Control travel control, estio, cluster or, and it is working. I guess that you should get this in your on on your on your machine.

A

A

Are people reaching to this step, I'm very fast, very, very low? Just let me know, let me give me feedback in case that I'm I'm fast or if you want to see things more, uh just water.

A

Okay, so thank you for thank you for the feedback. Okay, so I guess that you can see the application now you can play with the application right. So if you go to the left, you can change the clicker fatio. You can do some wall loss, some shop faster and you can see the type of traffic it doesn't matter, so you can play with it as described in the documentation.

A

This is a demo app that you can also to to use it for your needs and collaborate if in case that you want to add any features, so pull requests are, are open. Well, it's quite simple to work. Okay, so um I think that we have this step. um Okay, so let's go to chiari. Now we have this. uh The travel tutorial control the control uh service from the travel control name, space exposed through the english. Remember we have the book info, we have kiali and we have now these other applications exposed to the english.

A

So now uh chiari. What is going to tell me? So if I go to the travel control name express it's going to say: okay, this looks okay so, but if I select the other name spaces, it's going to tell me that hey, you are still some missing sidecars on on others, uh warlords.

A

This is quite normal and can be used for travel shooting, because this is a pretty common situation that you're going to start, deploying name, space and training applications and probably at some point, you may miss that somehow has not the proper sidecar for some reason right. Somehow it can give you this information from you.

A

If I go back to them to the graph. What's happened is that now the sidecar starts to collect telemetry right. It starts to collect a lot of information, and what gial is doing is to collect all these data and try to show you in a in a graphical way and a presentation, but I don't have. I have a mix of some of the walls have cycles.

A

Anki ali is going to grab what it can show you right in this case. It's just. I have some limited visibility of what the tower culture are.

A

Doing okay, so uh I guess that we finish the step number three: let's continue with the steps number four. I hope that everyone is following.

A

Okay, you say you say.

B

Look lucas there's one question: there's one question just about the demo itself when you're looking at the travel control page wondering where all those, where are all these initial numbers coming from.

A

B

Yeah, like the traffic, the the the things that we see right now,.

A

Good question yeah in the uh yeah. Thank you thank you for coming yeah in in the design of of this demo. Remember we have this ui, which are the shops right. This ui are simulators. They are ui, but no real ui. uh They are just emulators of traffic traffic generators. Let's just call it in that way, but this travel control is going to tell okay.

A

I want that the the uk shop is going to generate this ratio of traffic with this type of traffic, one traffic too, or with this characteristic and the same way for the italian shop or the other way for the friendship. This is what is doing so. Basically, we are simulating that because otherwise we will need to create a more conflict, but the idea is that we have a more rich number of requests coming from from everywhere, so this is different from book info in book info.

A

This is the the only uh sorry okay. I have a different okay in the book info to generate traffic. I need to go to the booking for page and refresh just to to really check that. Okay, I'm going to review one two right, so what we did for this demo is that that to create these emulators traffic generators that emulate this this.

A

This refresh the same thing that I do here manually, but from this business demo app, I can control the request ratio and I will see that you know that some shop is doing more requests than others. I can see in this europe map. Where are the new requests coming from? If I click in one specific, I will see some business information that okay, the requests that are that's coming for the to this city are coming from one shop or another. Just to to you know, to give me an idea.

A

Okay, for for this city, I receive a lot of traffic from the red one, which is a travel uk, some from the italian and the rest from the french. This is the the design of this.

A

Was this demo app so uh yeah? If any details, the the demo app is, is public and and the the code is is quite- is quite easy to follow. Okay,.

A

Okay, yeah, let me know if you have more questions. Thank you. Thank you for asking this. It's nice.

A

So, uh and- and you will tell me okay, uh do you stop a little bit in the missing cycle? Yeah, that's! That was on purpose, because I guess that when you're learning istio just trying to do in an incremental way, it's a muscle. It's a a logical step. We wanted to show you how kealikan can detect when something is, um is showing you information or is missing some cycles and can help you to travel shooting right.

A

This is a one of the of the features that kiyali is designed to give you to provide you this kind of information, but typically what we do. Typically, what we do is like in the book info, so what we do is to to enable the name, space and redeploy. When I have some experience with history, I say: okay, now this name space, I'm going to enroll the whole name space into the introduction. What I'm going to do is to uh to do by by labels right.

A

So I'm going to repeat that, so on this step, what we want to do is to label the the travel agency in the travel portal. Namespace.

A

Let me go to the okay. Let me close this.

A

Okay, uh this is wait. You know when I label the name space, the system workloads are on touch, so what I need to do is to uh delete the the word loss and redeploy as this demo. Application is safe to do that, so you can copy and paste the the commands to that.

A

And also when both on deploy, I'm also doing this on purpose. To show you how key ali now is going to see that something is broken, and I wanted to do to force some errors in the dashboard to show you in the in in the overview and show you in the graph hey at some some moment so someone some name space and deploy the warlords and and show some problems right.

A

So now I'm going to um to deploy the workloads here, I'm going to okay. This is my.

A

A

A

Okay, I written the chat that bart is commenting if kiali restart the posts.

A

Yes, one by one, if you label, as I did in the first step, I can do that by word by word, guys, just annotating when you annotate, the controller automatically keeps uh kubernetes, is going to update the pause and recreate the post properly uh to do that by the, but not if that is not happening just leveling on the the same feature: that's not happening, leveling, the name space. So this is something that is not yet automated for that to do that,.

A

Okay, I recreated the bots, and now what we are going to see is we are going. We still have the the console up because the the control application- I didn't, touch it, but you see that this reset.

A

Now, okay, so let me navigate first into the overview and you see now what is happening now if I filter by the travel. I hope you can see now you see more telemetry for for the name space.

A

uh If I some some basic controls initially, you will see that inky on on every page, you can select one or many name space right. This is one thing, and probably you will see that in most of the pages of pages you have time controls or refresh controls attend controls as we are collecting telemetry.

A

We can tell clearly what was the interval of the last interval of time that I want to get right. I want the telemetry for the last minute and it's okay, but if I take for the last 10 minutes, I will see that okay. At that moment, I didn't have completely telemetry and I have some problem for for some from some waters.

A

When I see some graphic also- and in this time contour, I can check the refresh interval, the refresh that keyal is going to to update the information or, if I'm doing something different, remember like editing a file, and I want to pause my my page for some reason. I can also to do that and to stop refreshing automatically into the key out right. This is quite basic on all the pages.

A

Okay. So, let's move on so we can see that all the warlords now have the recyclers to do that.

A

uh Now we are going to verify if the telemetry is updated and now, if we go to the graph, what we see now so if I click in the in the last 30 minutes, I'm going to see some errors because 30 minutes, I didn't, have all these walls in my mess. But if I click in the last minutes, I'm going to see a different, healthy telemetry right to do that. So this is what what I can show you. I can focus, for example, in the taber inc.

A

I may have different views, the application graph. I can show the service graph. I can show the person application graph. I can show the wall graph, okay, so but let's, let's try to start doing something more interesting or you know or something more one thing. He said. Okay, I can take a first view, but I want to check.

A

I want to query my graph. I want to extract more information, more useful information right, so, uh as I can tell, he only has several options in the display in the visualization in the display in display menu to visualization.

A

One of the interesting could be the frequent distribution. So you can see what is the uh the required distribution to all the words right. um One first thing that we can we can check as remember when, when I was explaining how the how the the the the travels demo works is that the the shop are requesting are simulating. That is a new user that this request. Okay, give me all the available destinations right, and this is going to do that to the to the other services.

A

What when I have hotels or free hotels, seat or or room right or the availability for the halters, um it shows all the hotels offerings in a first step and the next steps. What is going to? Okay, the users we simulated, the user, select quantity, remember this is what we see, but we what we see here, that one user can select luxembourg and then, in this case the the user is going to follow a next step and it's going to query flights, a specific offering for this hotel insurance, cars, etc.

A

Right, we can see.

A

That how these services can access to a discount services to check hey, I need to apply a discount, because you know this is a request that came from some kind of new offer and also how all these back-end services have a database, what they query. So we are going also to show how istio can work with.

A

Database, okay, so.

A

As I commented, one interesting thing is that I can show the the request distribution. I can see that I have more traffic for for the hotels, I have other options and if I click into the let me to show the service notes, because I want to show the service in this case. So in this case I have the award with the service. If I click into the hotels, I want to get more details of how the hotels are working right.

A

If I go to the side panel, I'm going to see stats about the telemetry for the inbound traffic. What's the traffic that is receiving all the outbound traffic, which is sending from this warlord to other services, but also we have a new tab which is called the traces.

A

These are the the the tracing that that that is, the cycle are are collecting, and now one thing that they can do is to overlap a particular trace, a particular request on top of the graph- and I can say okay, what a particular request is doing. So this request is coming to the travels and for this particular trace, it's calling the the hotels, the insulars and the cars, but not flies, because probably this juicer is is planning a travel just with without fly just by car right, and I can do the same by others.

A

I I can see what particular trace that I can overlap with my uh my scenario.

A

Another feature that we can show with with uh with kiali is that we can zoom on a particular application right and if, for example, I'm working in a in a large time space with a lot of services- and I want to study what is happening with the travis. I can double click in the in the tabs and just focus this one to see with more details. Okay and zoom into this.

A

And we can see how all the traffic is coming to to this version and it's moving, and it's confirming that. Okay, that, from the travis workload most of the requests go to the to the hotels one right.

A

If I use the the contextual menu from the from the workload, I can access to more details. I can double click and I can navigate into the details of the workload and in the details of the world. They can tell me what are the posts? What are the services that are in front of this, this workload, potential, easter config that affects only to the world, not to the service and, for example, I can see the summary of the traffic, the inbound traffic.

A

I can see the loss of this warlocks, you know destination and the sidecar I'm going to stop first into the the all about traffic and in the open traffic. I can see that, okay, that the most of the of the of the request, I can see that also to the to the hotel service, the old smallest one is okay, and I can. I can certainly do that.

A

Okay, one, as I commented at the uh in the graph, we can um combine the trace to a particular trace in the graph and to combine the telemetry, but also, if I go to the onboard metrics, this correlation between tracing and metrics are available in the chart. So this is the telemetry that that is collecting.

A

But if I click in the span duration, I can put together, you know into the the metrics telemetry, with the how the spans are working right and also I can move to them to the trace tab and in this three staff.

A

I'm collecting for this interval the threads that are happening, but what we want to do from kiali is to tell you, uh okay, um how to give you more information to tell if something is good or not or not, is good so and how we do that we try to correlate that with the east telemetry. So when we click in a particular trace.

A

What we do is to okay, how I know if a trace is okay or not it's okay, for the business. So this is something that we put this with a heat map of the existing telemetry, and we combine how a particular address is working with the summary of the telemetry for this interval, and we in, and it can give you more hints that, if how is happening, for example, this kind this this this is, this trace is more or less looking green. Looking good that compared with others, is in the same average of time.

A

I can go to the span details and see all the spans that are moving, and I can see also in the spans is this particular step of your business trace compared with the other medicine contests, are slope or or okay right. So this is something the idea is to to try to give you more information to interpret to profess all this low-level uh raw metrics that uh and sinus that history is collecting from the cycles in a way that you can consume it in a in a more business efficient way.

A

Let's see, what's what kia is doing.

A

Also for the locks, so in this case we are in the in the in the travis warlock and, as commented typically, I'm going to have my my application blocks right. This depends of of my developer's domain and how I developed that. But one thing that always I'm going to have in istio is the the htsi card and probably could be quite handy to see that okay, this is what my application is doing, but why this particular warlord is not working as expected.

A

What the istio is is doing or is not doing so, for example, in this case, I have also capabilities to to filter traffic. Let me to to see okay, for example. Let's me this city right. I want to show prague, okay, this city and ac okay from the application. I have a couple of travel quotes for this city, and I can see only I can see all the the communication that the psychiatrist is sending in low level to help me to understand our shooting. What is working and what is not working.

A

Okay, so I I think that tiadi has a lot of feature in in that sense, so the idea is that feel free to to review, but yourself this this this scenario. So, um okay, sorry, oh, oh, perhaps I miss the connect. I.

A

A

Okay, can you see again my my screen yeah, it's back now. Okay, thank you for some reason I stopped touching, so I think that I'm going to uh to give you uh five minutes to to continue doing the observing by your about your your your rhythm, your pace and before that we start with the step number five. Okay,.

A

Okay, so I guess that that more or less you, you read at this point that that you install travels demo app, you can expose the travel control.

A

You can start observing and and see what how the application is is working right and just to to see how qre collects all the low level telemetry low level tracing signals all level eastern configuration and is put into you in a way that can come help you to identify in a quick view how my application is working right.

A

Okay, so the next step is, this block is more to do to start browsing in the easter features, but just saying to these two features you see using chiari right. So this is what we are going to do. So at this point, so is you know uh everyone can work in your reason to do that, so I'm going to to follow these steps, but don't be afraid if you take more time or you want to to use to focus in one steps another for for the future.

A

So I guess that everything is very well documented and probably you can run this tutorial in this workshop at any time. Okay. So, let's start by a rigorous routine scenario. So now, let's, let's imagine that in my travel application I only have a travels version, one workload, and I would like to to simulate that I'm going to have other workloads and- and I'm going to prepare a scenario when I can show you how istio can help me to do dynamically routing requests to different workloads in a better smart way, a more complex or advanced way.

A

So we are going to deploy the two new versions, the travels.

A

Application: okay.

A

I'm going to the travel agency, I can.

A

Filter I deploy the travel dash versus to the traversing version three. Now, if I go to the graph, I will start seeing. You know that more.

A

Telemetry is something I want to show you. Let me to see: okay yeah.

A

I have new information, so I'm assuming in the in the in the travels app, but you know what what I see so now I have uh a travis app with three warlords version. One version two versus three: this probably is going to show me what is happening right and my for my travel shops again. Traversals are that are the the the walls that are generating this traffic, so this is transparent because it's the travel service which all the traffic is being ready, is being embarked by a travel service right.

A

But now, if, in this view, I unselect the service notes, what is going to happen that the service is advancing to the traffic to all the workloads from one warlock, I will see. Okay, I have some one third of the traffic to version. One version, two version three. So it's doing a wrong, drawing not balancing across everything right. So again, if I select the service notes, this is transparent from my application. This is one of the powerful of of kubernetes.

A

I can put a new wall load behind a service and all the water that are invoking these services is transparent right, but now I would like to show you in this scenario how istio can help me to tell a now, in this case, for testing purposes. I would like to to send all the traffic of the the french portal into one specific version of the the travels workload and the same for the english for the for the english portal and also in a similar way for the italian portal right to change this, this distribution.

A

So how to do that right? This is the situation that I have and now what I can do is to use, for example, to do that through the the applications uh either. So imagine this traffic, how I know that this stuff is coming from from this world versus others, so I can examine and still give me employees me the opportunity to examine the http traffic and build a routine high-level routine traffic depending on using the the the application headers.

A

This is what we are going to do in this scenario, so we are going to do the services one.

A

We are going to to to open service details and we want to create a newest request, rooting scenario right and in this sequential scenario, now we are going to create several rules in the request matching.

A

We want to say, okay, the portal that is coming from the let me to check the one from the travel 2k, the portal header, the request that that contains are http portal headers for the traverse uk and I'm going to add this match right, I'm going to route it 100, just only to the first workload and I'm going to add a rule right.

A

Let me do okay, this is one other rule, and now I want to create another rule, but for the other portal, the portal that is routing to the italian one, the big dot id.

A

I'm going to root two of the versions here, I'm I'm at a root, a rule for that and the second thing I, the portal, that is request, http, request to boyas.

A

If I can add a match- and I want to route it to the traverse versus 3.- and I add a rule to that- so you see uh this is the hideous portal. Let's start this travel. This is the request machine.

A

I'm going to send all this request matching to the television one, this other one, two server two and this other one to the third c, and I can create that so now what is going to happen right?

A

If now I select the service notes. Chiari can tell me: okay, you have in the travels you have now, uh let's say: okay, yeah. The graph is also detecting that I have a virtual service on this service right this one.

A

If I zoom here and I remove the service node, I'm going to see how now the traffic is moving and this is working and it's honoring, the the request match. So all the traffic from the italian portal is is is is being rude to the person too, and all the traffic for the for the english uh for the english portal is root to the version one and the same for the version three. But the interesting thing is that this is a transparent way, so I didn't change the application.

A

So this is something that the application is still calling the the service the same service and instill the sidecars are configured to do this smart routing to this workload. So imagine now that you have detected from the business perspective that hey, I need a special need for the for this shop right and this shop has a more traffic on one area, and I want to enable some new features that I want to test specific for this kind of customer right.

A

So this is one of the use cases where istio can come help me a developer perspective in a transparent way to add this smart routing features without modifying anything in my application, in a in a transparent way, and also to show you how I can do that from from chiari right in this case the travel services. What it's doing is. Okay, I generated the the htc for you and I can show you the channel that I generate. This is the the the the resulting jammer.

A

So it's just doing the the rules with the matching for travel, uk and just updating the way for you and thing that also can can help you not only the built-up services but also destination rules. In this case, it's just mapping to the world laws, but can tell me I can see the generated easter config that I would need to to elaborate or to implement this use case right.

A

Okay, so this is the for. You can also modify and play with these scenarios and try to to to change more ah yeah, fantastic metrics.

A

If I now check the e-moment for this service, no sorry for the wall loss. What I wanted to show you.

A

I wanted to show you if I okay, I can select by.

A

Remote applications- and I can see that at some point I had all of them, but when I enabled at the point that I enabled them the easter comfy for this, uh the traveler's version, one workload is when I can confirm that at this moment right so the others, this warlord is receiving traffic only for the for the okay, the english travel shop, okay,.

A

Okay, when you are done with this scenario, uh I think that we can update or test you can test offline, other configuration just to get familiar, or when you are done, we can delete the configuration so to delete the configuration that are linked with a service.

A

Yes, we can click here in the in the the same menu of the with us. We can update it the existing one. I can change that or, for example, I can delete it for for this scenario, and I will remove the this. Is the conflict for this element and we will see how we go back to the other.

A

A

Okay, so yeah thank you, for uh we are going to test a new, uh a new use case for easter, which is a full intro. The fall injections uh comments, so I saw some comment in the chat that some folk was telling about introducing cars. Monkey could be sorry, I I I know that jay is reading all the the comments about so monkey testing. Something like that. So this this is going to be in the in this scenario, is going to be in that in in that line.

A

So, as we show in the at the beginning, so is that, okay, that we identify that the hotel service has a lot of traffic, so it has a lot of load compared with others right so also, kiali can told me about the response time.

A

Okay, I can see okay, the response thing, as this is a demo application and we have a very powerful cluster supply provided by by google. Typically, it's going to be okay. If I'm in my laptop, probably, I cannot run for notes of virtual machines and then the response time could be more slower that I can see here so this.

A

This is the streams, are pretty good, so the load is is quite good so, but I would like to you know: at the beginning of this workshop, we detected that this author service was having extra load, so I was having okay, I'm doing the a query for all available destinations and I'm and after they give you the user. I just getting a second query to that.

A

So one scenario that we want to test is that I can introduce uh errors on purpose and I can't try to say: okay, I can introduce some kind of of errors with statistical probability or I can introduce delays uh why I should do that also to test how resilience or how stable my system could be for in this scenario. So this this, even if this demo is, is quite small, but you know I have several services. I have several award and what's happening, one of them is failing or is slower than other.

A

How is going to impact that to the other? So this is what we are going to test in this scenario: how to do that? Creating a uh introducing a four injection scenario: how to do that? How can he ali help me to to perform this scenario, so I can go to the holder service, which is the service that I want to test.

A

I can click to the actions and I can say hey. I want to elaborate a full injection scenario.

A

I can choose a delay on a board, but in this case I'm going to work with the delay, um I'm going to delay 100 of all the requests that I received with five seconds right. So, let's, let's do that?

A

That's it so at this moment, so the mess know that every request address to the hotel service should delay by five seconds right. So we are going to see how that is impacted into the.

A

A

Okay, we see that we have updated. The kia is telling me that now I have the hotel service here.

A

Okay and now it's something that is, if you see, if you have to display the response sign, you are going to see that the things that the edge has changed with information. So I would like to show you uh an interesting feature that, okay, when I have a a populated graph, how I can see how.

A

How some wallet is slow or how I can detect or see how the the slow is propagated? So this is one the graph has a powerful feature which is called fine and height. So how to use that. So it's going to allow me to ask me to to to perform some kind of queries. Let's call it queries, it's not really a query, but it's going to filter some elements. For example, I can filter the edges of the graph by, for example, the response time and I'm going to filter.

A

Spawn side, which is more than one second right, oh response time, one second.

A

And it might not know to show that uh now I guess that you can get an idea of how the impact of a slow warlord can affect the other's world all right. So I can query the graph the clearly the key holy graph to to show that okay, a slop in the in the in the hotel world, it's okay, everything is green, so nothing is failing, because I have a very powerful. um I have a very powerful cluster, but I can see that how is propagating these latencies into the the system? I see that.

A

Okay, all the the portals, has impacted these versions. Okay, so uh let's, let's take a uh uh a more close how to get the details of this metric. So I guess that I can see into the this is also a way that I can show them.

A

Okay, I can show that the concept of software destination world, when I get when I collect metric when the istio collect matrix right in a scenario like this, that okay, the travel service, is calling to the hotel service.

A

So it's just collecting telemetry from from both sides. It's telling from the travel service. This is telling to the to the telemetry okay. It takes this time to access to to query the hottest one and from the author's service is also okay. I received this query and it took me to do that right, so the telemetry can give me additional information on how the cycles are working, and this is what we call the report.

A

So if I go to the inbound matrix depending on the source, if I select the reported from source, what I'm going to see here is how is the telemetry from those services that are invoking the hotels so from this side of the of the communication?

A

The travel services when the travel services is is calling the hotel services how much it takes for for that, and this is what what is telling in this case, so that in this case, we see, okay, that we have the five seconds delay with some cases where, where the delay could be even the table now, this is the percentiles that I can also change here. The quantiles I can just to focus on the average.

A

The average is five seconds, but I can check that there are some very slow queries that may happen right, but from the from the the the hottest world is, is working okay. So what is happening all right? The telemetry is telling me that okay, I'm just responding the warlock, is working fine, but what is happening here right? Okay, so what is happening here is a feature of them by proxy right that the the proxy the cycle that is injecting this, this foul injection is the proxy which is located collocated to the travels world.

A

So this proxy is telling okay this this service, the travel services, is invoking the the hotels I do the request, and when I receive the request I be injecting at the site, then the delay. So this is something that is doing. This is what I have different information, depending of which report that I'm using to to take that.

A

Right: okay, I guess that we have reviewed all of this. So again, uh these scenarios are quite interesting, quite useful to you know to to check the resilience of application to simulate um problems in the system. This is quite typical to say: okay, everything is green, so the word laws are still working, but how I can visualize the impact of the delay, because my my network, this is one of the of the interesting features that really can can make for you right.

A

Okay, when you are done with this scenario, I can go to the to the hottest one and I can delete the easter configuration as I promise you, I won't use jammer files to to operate with istio. This is something that I would like to show you with these use cases, and the next thing I would like to see another quite typical use case, which is the traffic shifting right.

A

Okay, so the traffic system is, uh it's also a very common scenario or av testing or some variation about the av testing scenario. So let's say that that I have uh an application. For example, like the travis application, I deploy several versions.

A

The version one is the old one, the consolidated one and the new version, two version two uh so version two or versus three are version that I want to test, and I want to to evaluate which one is working better right and a quite easy way. So a traffic system is a way to tell kiali. I go to the travel service. Click in 2016.

A

Sorry that I want to to keep sending the 80- or you know, 60 of my traffic to the version one, but I want to test uh something like the 20 between version 2 version c, because I want to check dual weight with which of these two versions are working better with all the traffic right.

A

I can generate that, so this is going to create some rules and beautiful roles with with basic ways. It's got simple scenario and in the graph now I'm going to show the service. This is the same, but if I now select the required distribution for the travel service.

A

I will see how remember the 60 percent of the traffic is expected that is going to to came to the to the mutual one and the other person dates are going to come to the person to verify. So what what is my goal.

B

A

What I would like to do to do is to to answer, if which of version two version, three are better right, are working better or if there are some significant version between these two versions to to change to one to another.

A

So one thing that that is interesting is that uh in reality we bring the concept, the logical confidence application, so an application in kubernetes is not a an entity. It's not.

A

An element: it's basically a logical group using application levels, but okay of a set of walls and services that are grouped logically, but why this is important for from istio perspective, because you know the telemetry is also grouped by application and that particular grouping can help you in a scenario like this, that I have several workloads for these applications and in my metrics I can see. Okay now you are showing metric group by remote application. But now I want to change by the local one I can test in my settings.

A

I can, for example, do the virtual one is: is the one that I know how it's working? I can select this one and I can just to compare how beautiful two and different c I are working, which one is working better or or if I need to select, if I'm going to do a rollout completely to one or another right compared with the whether it do the league or inner one.

A

I can go to put back the the version, one, which is the the whole version, and I can you know to do comparison and to try to respond which one could be work better in my in my salary, okay, so with a traffic.

A

A

Okay, I think that we show all the information.

A

So once I finish with this this scenario, I can go to the actions to delete the traffic shift, team and yeah. Let's move to to the next scenario, to the tcp.

A

A

So in this new scenario, I would like to show you how istio can help you with other services, non-typical services, uh pure tcp services, like, for example, the database right. So I want to simulate the the following. Imagine that you have an application and you have detected the double agency, one.

A

That my uh my database uh is, it may have something in my data that I want to test. You go to your database admin and your database. Admin is proposing you a probably.

A

We have some tables with without proper indexing, and I would like to test you know which indexing could be work better, because it's something we are more in distributed scenario, and I would like to to to evaluate two potential alternatives. So this scenario is going to show you how I can deploy also to warlords for database.

A

Okay, I'm going to.

A

A

I'm going to to deploy to two new walls for the mysql database.

A

I'm going to check when the deployments are fully routed.

A

Okay, it seems.

A

A

Okay, so now I see that I start to see how regrets are balanced between.

A

A

Okay, now I have the version three right, so the the same, every testing that I I was applying to my application. I can do that with the database. In this case, I may have a similar scenario I would like to. This is the version one was the neolina database scheme right and I can define into the mysqldb a tcp traffic shifting where I would like to put the 60 of the traffic.

A

Into the original version, the version one and I can put the twenty percent of the other traffic to the other one.

A

Oops, I have some problem. I don't know why.

A

Now I can see this one's.

A

A

Okay on this case, as this is not http protocol, I guess that I only have the request read matrix defining here and I will see how it distributing between uh the dcp traffic method that is collecting by you is distributing for version, one between version, two or person, three.

A

If I go to the service my secure.

A

A

I can also filter by the local versions.

A

This is the version one how this is working and I can compare the person to ambassador about.

A

A

Okay see so on on this case. For this scenario, probably I don't have big significant change between one another and probably any of of two indexes policies. Could work really well for for us.

B

A

Okay, so um if you are done with the gcp traffic shifting scenario, so we can just to go autocliali to remove the generator configuration and, let's move to the next show calls.

A

So the next element that we are going to do is the.

A

The request time app right so uh remember the the first full injection scenario that we were working on. So we were introducing some delay into the hotels and we put in five seconds delay into the into the hotel service and yeah. We showed that, okay, we we we slowed down all the systems right, but um but everything was what was green was okay right on this scenario.

A

We want to to show how we can introduce uh and we can just to stop this kind of delays and in so in some cases we can apply some secret breaking techniques into our application. It means that probably some warlords may prefer to to fail and recover on a slow situation, rather than appropriate, a very slow request across all the workloads, because it can impact uh in a in a worse way the rest of my application.

A

Instead to say the world hey, this is pretty slow, so I will fail and- and I can just to recover faster- that if I have a very healthy path, slow request right. So this is what we are going to this scenario. So what we want to do in the first step is to repeat again the the default injection that we introduced in the hotel.

A

Service, so we go to the hotel service. We go to the actions. We introduce a delay here about five seconds right. So what is going to to happen is that if, if we go to the graph- and I show the response times, the hotel services is going to start to to run quite a slope and it's going to to send five seconds as average or even the double it's on request for everyone. That is invoking this like this and this slowness, as uh as we show on this case.

A

Okay, if I query.

A

I will see how all the wallets are impacted by that, but in this scenario, what we want is to okay, so uh one particular portal cannot affect a a delay for more than two seconds. In that case, this workload, the voyas portal in this in our sample, prefer to fail and recover and to send to the user that the application is not available instead to propagate uh five or ten seconds delay across all assistance. So we can do that also with istio and with our on the business rule.

A

So let me to show this how to to to perform this scenario using er right. So uh as the service that is, we are going to target, is the the travels one?

A

We click into the details and we are going to again to elaborate our request routine slide right on this case, we are going to select the the portal header.

A

We are going to um to create a match to the portal here to the bojas.

A

And in this case it doesn't matter what we are going to say: we are going to root to all the world laws, but what we want to do in this case is to add a delay of a timeout of two seconds to add this rule right to request matching to all the that you guys coming to portal to buy us okay route to um to all the travels with a request timeout of two seconds- and I add this rule right next. I need a default rules with requests for the any other request.

A

So because you know the builder service is working with a list of rules and the first rule that it matches so it's going to apply, but if it doesn't have any rule any default rule you don't know what is going to happen right. So this is what I is always the best practices to add in in individual services a default rule for any other request in case. That's that you have some matching case that is not taking.

A

So I'm going to create that, so what is going to happen now, so the expectation is that.

A

Now what I expect is that uh the the slow requests are is going to be okay for the italian and the english portal, but at some point the french portal is going to receive. Fails, I'm going to say no! This is not okay! Okay! Now it's failing right now. You can see that this is in red.

A

It's getting some problems, so let's go to this detail, I guess okay! Now I can see you on the edge. If I click on the search, I can see the error. If I click on the edge, I can see a flag and the flag is telling me okay. Now I'm going to put in pause to something better than frost. I can click here in the in the flash um in the edge and you can see in the in the flag tab. How keal is telling me that this this error is happening right.

A

I have a uh an error connection, because the downstream connection combination right. So, if I go to the.

A

To this application to this.

A

To this wall load- and I try to do the settings by, for example, for the response code- I can see how now I started to this was when, when I get my my all requests, but now I have the new request no longer than two seconds, but with error. What does that mean that you know that the istio is telling me okay, so for some workloads this um this delay is.

A

Let me do okay to show a game. This is scenario, okay, so, okay for some wireless, I can accept as as a slow uh as love request, but for this particular um warlord it cannot be good and I prefer to fail sooner to apply back pressure rather to to propagate as a slow request that can impact in a walkway in the rest of the system.

A

A

Okay, I can see more details. I can just to to get all the edits in the trace into the the metrics so uh feel free to to change the scenario to this new new thing. You know all the steps are quite straightforward. When you are done, we can, for example, to go to them.

A

And delete the route you need, the configuration.

A

A

A

Okay, okay, so let's go to now that we are, you know, so I see a lot of activity on the chat, but I cannot read sorry so just you just talk me if, if I'm I'm running very fast or very slow, you want me to to to clarify some topic so in in these scenarios, as we are starting to introduce uh errors, we are starting to do resilience, technical retrieve techniques on time of techniques.

A

Just to to you know, to tell the istio that the the voice, I guess, can stop propagating errors and can like I can apply sequence breakers techniques right to to fail sooner rather than to to propagate some errors, like my sister right. So uh as you know, we we have a a pretty simple demo.

A

So now we are going to to work in our scenario in a secret break in a scenario a little bit more complex right, and I want to show how I can you know to to improve these techniques and and I'm going to apply a real secret breaker scenario she can break. It is okay. At some point, I want to close some connections, but I want to allow other connections coming into my scenario right, so how to start this this example, so we are going to deploy a new warlord in the teleporter namespace.

A

Which is a more intense, a more heavy road testing workload, all right so now this deploy, we are going to check if the workload is.

A

Okay, now, let's supply.

A

We are going to to see.

A

Okay, so now I have the.

A

The service graph I have the wall.

A

Okay, and let me show you the request.

A

Plate, okay, now this is the view that I wanted to do, because I have several versions and now I have a lot tested right and if you compare, if you by application group by application right so remember so I have several addressing. But now I'm grouping my applications.

A

So this new wallop from the teleporter is just sending 10 times more traffic than the all the three worlds together. So it's just sending overload to the travel science to do that right. So uh you know I need my google cloud platform and it's working well, because I have four nodes, but probably this scenario could give you some problems in your local, your local machine right, and what we would like to do is to tell something that hey I don't want to.

A

I want to slow down this is this scenario. I want to apply a circuit, breaker and to say hey. I cannot to get all this uh this traffic. I need to limit this capacity. So if I go to the traverse applications- and I see the inbound metrics, I will see the difference so this.

A

Let me to show that the last minute- okay, so this.

B

A

The request from the load tester, so it's a lot of traffic compared with the traffic that is sending these other shops. So this is what I'm showing that now what I would like how I can limit this and how I can just supply secret brake techniques to do that right. This is what we are going to show in this in this new scenario. So again, I can go to the to the travel service in kiyali and I can see okay.

A

A cq breaker is something that I can apply to any scenario I'm going to do a traffic shifting as it is, and if I show the advanced options.

A

A

I can go to the tab to seek a brief tap and I can add some kinds of settings short advanced settings in the destination rules to the connection pool the maximum connections that a whole lot can handle. Concurrently, I can define what is the cue of of pending requests to allow and I can also to define you know how I can modify the consecutive errors before the driver can be.

A

You know, put it outside of the of the wrong rolling leaves of the the side, part the walls of the wrong drawing of the load balancer for this scenario, so this is something that they can do from kiali from the any of the wizards go to the advanced options. Remember at the beginning of the warsaw we created a um we created a a gateway. Now we are going to sell a car, a secret bricket and we can apply to the travis one.

A

Okay, so uh what we should expect that at some point?

A

Okay, so let me first to show you what keali has generated hacked it first virtual service to distribute the traffic in that. But remember all the details or the secret traffic features are applied to the traffic policy. One.

A

Okay, so let's move to.

A

The do next step, so if I go to the load.

A

Balancer, I'm going to see something like like this. Okay, that I have some you know some requests that are failing with the flags that okay, the absolute overflows the secret breakers, is open. This is failing with a it's an error, because the secret breaker is applied into the into the proxy, but others are going to work. Okay right.

A

If I show this into the automatics okay,.

A

I think that I'm going to show this from the okay from the load tester, the album metrics.

A

No from the source, okay, I need to to check by no not my very most question by response code.

A

Okay, so I can see how it's working, so I reduce the pressure, so I'm just allowing.

A

The the request that one, that my connection pull settings are working, but those requests that these connection pull settings cannot get for. This particular warlord are just returned by error by the cycle, so this is an example of how the the secret break is.

A

Working okay, so we can just do when we are finished. With this scenario, we can to.

A

A

To the titles one I can delete traffic.

A

A

Okay, so the last step of this of this block of of the connectivity is the military one.

A

So the mirroring is a particular a scenario similar to a traffic system, but I can just copy a request of the of of the the services and to to route to a mirror workload instead to to put it out of the ban.

A

A

You can go to the services one and I can go to the travel one and at some point I can just to repeat the graphic shifting- and I can say okay, so imagine that uh that I want to to to consider that the version three of the traverse is pure experimental, okay, and I want to take okay, how I can to test this this world or this pure experimental workload in a non-intrusive way in my system right.

A

So it's one pattern is to put it like a mirror right clicking in the mirror icon, and I can tell to the si cards. Okay now I want to distribute just 50 of the total of the of the on the official request. Let's call this in that way.

A

Okay, four front travel, one for tremendous one time version two, but I want to mirror fifty percent of the traffic. You know to the person three for testing a pure experimental way that this request won't be back due to the color to that.

A

So this is what what I can create here right at some point in the time, so this travel- if I show the version of of the travels now this one with the.

A

A

Okay, let me to put.

A

A

I apologize because I realized that I didn't enable the traffic animation at the beginning of the warsaw, which is something that I should do you know. One of the of the of the nice feature of of kiali you know is, is now that I can to simulate the same variety of the traffic of my edges can be also been animated right. So now what what? What I define is that this wallet is going to be mirror. That means that the official traffic is going to be.

A

uh Let me to distribute it between this and this order, and this traffic is going just to just to be mirrored to get.

A

Let me save that. Okay,.

A

How I can just to to show that the the all the traffic is going to be rude by version one and version two, but I will use a submission. Three, like a non-intrusive mirror, I'm going to copy uh 50 of the requests to this other mirror for testing purposes.

A

Okay, when, when we are finished, we we can just to delete the configuration. So I know that we are late. We are just finishing the.

A

The complete tutorial, so with this step we are more or less finishing. The the connect block is uh a long block. um I would like also to to finish the tutorial with the as an example security. So for this last scenario, I would like to show you how powerful can be to combine the the all the telemetry information that is collecting with security with security access. That kiali can provide sorry that the issue can provide an ankylic and can come by to generate automatically configuration.

A

So one thing that we are going to do is to delete the load tester. So just go to your your.

B

A

I'm going to to delete this this tutorial sorry this uh this word, the last tensor warlock.

A

And in some time this traffic is going to.

A

A

Okay, why this not? Oh, okay, it's okay! So I guess okay! Now! Okay! So probably, if you follow me, um the love balancer warlord has disappeared right, but now I would like to show you one of the features. I guess that in in other insta conference, they they showed that in in oh, I can show okay. This is what I wanted to do. Sorry, uh I want to show you how I can enable the security flux and by defaulting in istio, is securing all the connectivity about that right.

A

um One of the interesting features of security is that is to allow you to define practically anything about security right.

A

One of the feature that chiali allows to you is to say at some point.

A

I would like to look to the current traffic: take a snapshot of all the current traffic and secure all that all the traffic with alteration with authorization policy rule so how to do that from kiari. So I can go to the overview and, for example, auto name. Space. Hey now generate the traffic policies for me.

A

So if I click here, what kill is going to do.

A

Is to first generate all this history configuration first to create a deny or rule by default, so anything by default is not allowed to do that and I'm going to allow just case by case specific specific words. Okay, so as the previous traffic is going to work as usual right. But now, if I try to to redeploy them the lot tester.

A

Probably, as this new warlord is not allowing to access to my mess, I will expect that is going to uh to show okay. Let me to see if.

A

Okay supply and now.

A

Okay, what I expect is that, okay, now I can see so now, can you see the this um this load tester? So it's failing everything. Why?

A

Because, in the previous step, I set up all these security authorization policies for my travel int based on the current telemetry, but now when I perform this step, the load tester work, a lot warlord was not part of my regular traffic, so this is a way that that can show you how really can easily to to help me to draft a more complex security uh scenario that I can work to do that right, okay and yeah. I guess that's uh almost is so um and and yeah.

A

So if you are here, congratulations I think we are close to to finish the the walls off. If you could follow all the steps, I I hope that they and me we.

A

To show you how to understand what is doing you know how to not only see you know the the static part of istio you know, but also how can kealy can help you to generate configuration? How help you to operate with the with your issue configuration a high level scenario, how we can just to provide high-level use cases driving with the most common operation that you would perform with the istio service mess and yeah?

A

So if you couldn't follow all the steps remember, all this documentation is is public as it's available to you, you can perform run it offline, um as I repeated it at the beginning. We are in days room and if you are an old developer like us, we are also using irc.

A

Okay. Thank you.

A