Kong Tech Talks, 28 Mar 2023

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: A Developer's Guide #KongKonnect: Your Ultimate API Control Plane 🌐

Description

Join @ViktorGamov Principal Developer Advocate at Kong, in this hands-on, comprehensive guide, we will dive into the power of Kong Konnect, the managed control plane for Kong Gateway designed for seamless API management.

We explore key features for API management, such as rate limiting based on consumer groups, API authentication using API keys, publishing APIs to the developer portal, and enabling dynamic application registration so developers immediately start consuming your APIs.

Plus show how to leverage API automation to streamline your workflows.

This engaging, practical video will revolutionize your API strategy and help you elevate your services and applications to new heights! 💪 #KongKonnect #konggateway

Subscribe to our channel for more API management content and visit our website to learn more about Kong Gateway: https://konghq.com/products/kong-konnect/register

A

A

We're gonna talk about developers guide to multi-cloud API operations with Quant, connect uh very mouthful uh and we're going to be covering a lot of things um for those of you who first time join us for Tech talk. My name is Victor gamof I work here as a developer of the kit with Quan, focusing on you know everything uh starting from Gateway to uh service mesh to um talking about, like microservices developers and and obviously Quant connect with which gonna be focused on today.

A

Let me know in the comments: uh what are your expectations of here so you're a developer, so it's it was tainted for developers and I'm, trying to explain a little bit on operation side of things and how we operate in gateways and how the people operate in gateways, and why um why you need you know things like connect, so that's my quick agenda um things we can.

A

We can have a multiple turns depend on the questions that might have, but essentially I would like to talk a little bit about the why developers should care about connecting or like if you're, not aware about connect. uh Why are you sleeping on this? Why are you missing out with the um learning about connect, um we're gonna, we're gonna, use very exciting technology to explore today?

A

So um I was very inspired by internal presentation um that my team were doing uh with some of the chat, gbts and open AI apis, so we're gonna look into uh building blocks of this chat.

A

Topt we're going to look into text completion API, how we can use it and how we can turn it into something that will be consumable within organization, something that a Quant connect platform will allow to configure and we're gonna play around with the um the the Gateway side of things and uh we're gonna cover some of the basic use cases like defining a routes defining a some of the like simple, simple, plugins, to more advanced uh rate, limiting capabilities based on consumer groups. How about that? This is very inspirational slide.

A

There's a lot of things going on here, and this is something that I will show in the very beginning to give you kind of like a road map where we're going with all this and we've different components that cover different aspects of your API management. Journey specifically, uh when you see Everyone likes to build the platforms, uh but the because platforms are great, they cover all use cases and sometimes developers are to kind of like get lost in in this kind of like uh the architectural vs diagrams.

A

um Today, we're going to be focusing on Gateway side of things um and uh how we can provision how we can enable the plugins configure plugins, so we're going to focus on a data plane what we call like a the Gateway setting for API management, um how we can control and configure and at the end, I also show how those things can be replicated through automation. Automation is important, um and that's all that's that's it. That's that's all my slides before I join with connect.

A

I will like to give you intro into how conch Gateway essentially deployed when you are a developer. You really want to care about two simple things. One thing is your con Gateway instance and configuration of your Quant Gateway that will include your services, your routes, your plugins, simplest, possible way. How you can deploy your developers deploy Quant is by defining this declarative configuration. Enable quonc is what we called a DB less mode, because this mode doesn't have any database here.

A

That's my ribbon. That will be explaining what happens on the slide. So, with this in clock, declarative configuration, we will have a Services configured. We have a routes configured and plugins configured. If we need to update this, we need to submit this new file to config admin.

A

Api rest of the admin API and points will be available for read-only, so this is kind of like an immutable immutable environment suitable for development suitable for integration, testing and suitable for some of the corner cases that we developed inside the Quant and we're using this, for example, in Quant Ingress controller, which our, which is our a management layer for kubernetes. If you, if you use this, use a thumbs up thumbs up in in slide, if using the quantity grass controller right now now, bayani developers mainly operators.

A

Many people like to use our admin API to configure Services, because now think about this. With this yaml, you have only one file. If you need to just change something in the service, you need to resubmit. This configuration you need to update this, so um it would be great if you will have admin API that allows you to do this incrementally. So that's why we're running Quant in our uh so-called a traditional mode, so with traditional mode we introduce a database database will have a persistence layer for our API.

A

In our configurations, all these things will be stored inside database and admin. Api will enable access to those Services through admin API. So you will be able to do mutations. You will be able to create service, delete service and after restart, you will be able to configure rehydrate this configuration more importantly, you will be able to use multiple instances of con Gateway with the same configuration, even though you still need to have.

A

This is kind of like a clustered mode, but instead, if you need to like Route traffic, you still need to have a load balancer in front. You might even have you know the. If you go in the fully meta, we can have a Quant Gateway in front that will be configured with the Kong yaml file and after that we'll be load balancer between those those nodes. However, this is pretty cool. This is something that many people are doing. This is something that uh many.

A

uh This is that's why it's called traditional deployment um and this traditional deployment can be extended by introducing a few more components. uh We do have a con manager for a UI and for managing things, and they can either use same database uh developer portal for publishing these services in enabling the services for for other users, and now we can start seeing uh dependency on this database database becoming a bottleneck and because our gateways needs to go there and get configuration if configuration was changed through the Quant manager.

A

With this database, there will be some time a little bit of time that happens between the changes will be pulled back into configuration, so there would be some of the eventual consistency. That's for my distributed system, folks, eventual consistency for configuration on the Quant now- and this is the moment where, a couple years ago we turned things around. We turn things around and introduce this configure duration that we called hybrid mode in the hybrid mode. Quant Gateway is not connected to this database.

A

It's not connected to database, but it's connected to another con instance that will have different role. We separated the role of data plane that will be responsible for pushing you, you're, pushing traffic through your data plane and your control plane.

A

It's also another instance of conch that will be responsible for updating configuration, admin, API syncing database when you change something on control, plane, control, plane, will push it down to different environments, drastically increases complexity of the system, but increases uh developer productivity because you don't need to worry as a developer, you deploy your data plane, connect to your control plane. You get configuration and you're good to go, and your Quant manager your developer portal. All the things will work.

A

Someone in the comments might ask Victor, but what, if I want to have configurations um that will be depend on environment, say I do have my testing environment for con and I. Have my staging environment? I do have my QA environment. I do have my pre-production production environment. How I can do this so, with this configuration things becoming slightly easier than it would be with the previous use cases, because um you need to you know, replicate this database. However, you still have this um situation when you need to migrate.

A

This configuration between business environments with declarative configuration in tools like um like DAC, you will be able to say, create this environment in one place. You can dump this configuration after that propagate this configuration to another environment. However, with the hybrid mode you need to manage your environments, say QA, you need to basically replicate all this thing into different environments, and this is where uh connect comes into play and solving some of the some of the pains for developers where so, first of all, you don't need to maintain database.

A

We maintain a database, we're managing the configuration layer and, through this uh fantastic feature, I think it's a one of the Revolution features that our customers are loving and is to have a runtime groups where you can segregate different environments. So in this case it can be Quant data plane that runs on your QA, and this one will be runs on your prod. More importantly, with this configuration you can also scale this environment not only within um different uh different environments, Dimension, but also in geographical dimensions. You we support a geo-distributed deployment of the content.

A

You can segregate them based on geographies, because you know, as the developers will like to build our stuff. However, uh the boring things like governments and the borders and local regulations come into play. So we need to obey those things. When we develop an application, we develop an application in the U.S, but if we're processing data of users who are in in European union, we need to store this data in European Union. So that's why this comes into play as well.

A

So what you need to do without connect, you probably will go, and you know deploy this in somewhere in in the European Union with connect. You just need to enable another geography and uh you're good to go. So as of now, we we have a fully managed a control plane that we're going to be focusing today, so control plane, we will be operating against control, plane and control play will be responsible for shipping our configuration to our data plane.

A

Let me know in the comments if, because this is the point where I'm switching the gears in going full in code and demonstrations and all these things- if something I just said, is not quite clear or you want to have some sort of clarification about these architectures in general. Again, you need to remember, which have only um four four deployments: dbls mode, simple as possible for development, and- and in this case your clone gamble will be your source of Truth with traditional mode. You have a database, you have admin API, you can configure Services.

A

You can have multiple gateways connected to the same database. They can share configuration one to one and one of the gateways can change this configuration. Another Gateway will pick up this eventually with hybrid mode. We have a control plane that will be responsible for consistent configuration and pushing back all this configuration to um to our data planes and with connect. We have a multi-attended, fully decentralized, fully managed control plane that allows you configure not only different environments, but you can have a like really um um not sophisticated, but really complex deployments like a really heterogeneous.

A

You can have a deployment in Docker and and in kubernetes, and they can be the same control uh same runtime group and same configuration will be shared. Yes, John Williams asking the Fantastic question. So when we're writing this in the hybrid mode, how we can manage a multiple environments. So in this particular case you need to replicate deployment, maybe smaller scale, uh but still you need to replicate the deployment across multiple different environments. So for QA you will have your own installation of control plane for QA database.

A

You need to have a QA database to store the cone and obviously you need to have a QA.

A

um Create the the gateways you can use if you're running this on-prem, you still can use um something that we called workspaces to segregate the services, but um the the thing like runtime groups you can get as of now uh only in in Connect. Now, let's dive in into this demo, so I have a battery comment to clarify in hybrid mode data plane does not have DB and mass connected to control plane, absolutely right. So with the hybrid mode.

A

um One of the processes that Quant binary can take on is either data plane, control, plane. You can. You can configure this, and only thing that Quant Gateway in this particular case will require is connection to control plane and it is. It is not uh like it is bi-directional. So when the con Gateway starts, it has a configuration where to connect for control plane and how to connect this and the rest of this stuff.

A

This this thing will be constantly updating, is bi-directional in hybrid mode actually think about this, because the current Gateway doesn't store anything. So there's no database there's nothing. It is very good um scenario, for um you know, for security reasons, also with security reasons. Like if your gateway itself would be compromised, no one else will be able to access um your uh your your configs and because you know, they're, simply not storing and uh only way to to connect to control plane would be.

A

You know using um key and certificate to connect, because the connection between control plane data plane is also secured using the SSH. Yes, thank you for this question uh Patrick. What I would like to show you today so I'm gonna start with with simple cluster I will be running. My Quan Gateway um I will be writing this in um Us East.

A

I guess it's US1 or ES4 I will be running this in kubernetes, just simply because I want it. If I will be talking um about the uh that's my attempt to do kubernetes stuff- maybe it's not right. I need to I need to play around with my drawings.

A

um This is available right now. It's it's on public internet. So, if you want to follow along, um you will be able to do this um before I. Do this, let me show you what we're going to be working with so I like to use a public API that is available and we're going to be using this um text completions, how many of you played around with charger PT like you asked to write a email for you, or maybe asking questions and trying to convince it that that's um it's wrong on something.

A

um So essentially it's done through this, uh this framework of text, completion and the when you submit the prompt um it will use the information that it trained on to provide the response, and they they expose this as a rest, API. There is a rest API that you can play around, and this is what we're going to be using today, so I'm going to be using I wanted to to provide access to this thing to my developers inside my organization.

A

However, I don't want them to register on this thing have um like put their credit card, we're doing the centralized way so I already provide this. Only thing that they will need is to connect to my to my Gateway and I will be able to control who can see which, which things since I mentioned pricing.

A

It's very important, because in the past, um in the past we'd like to talk about when we're talking about performance, we like to talk about things like um optimized algorithms, maybe faster networks like putting the faster hard drives and faster CPUs these days.

A

If you want to have a optimized performance, you just need to swipe credit card much harder because you will get a say more powerful instances if you're running this in cloud in the things that you're running in AWS, for example uh same here with um with chat GPT, we have different pricing based on uh the quality of model of the the model that will be responsible for generating text.

A

Let me show you how the quality of life changes if we're changing different models, so we're going into this um that I'm using this chat, GPT completions URL, so uh we'll be trying to do different models uh again start with this simple model called um Ada and see how this will be able to work, and here I would like to ask what is Chrome and when I submit this question response. Time is uh 900 milliseconds, so it's relatively fast, but we're getting uh interesting response here.

A

It's fast, but response would be calling this computer game played by zones on the twin planet player. Armored Kong is computer. Gameplay zones, tattooing plane interesting. So if we're gonna be say like okay, if I want to pay more for better model, uh let's see. There's another model called Babbage um and when I run this and a Babbage response time, I guess I guess some caching happened.

A

um It says Kong is a character and the mascot of Nintendo Entertainment System, okay, but not exactly what we expected right. So, let's go with the more advanced model like a call to Curie all right. Quonk is a software company that creates video games and other software products. It closed to uh that's a software company and other things, but it's not that close with the uh Mario and Donkey Kong games. So now, let's use DaVinci, which is uh more advanced model, more um more sophisticated model and this model used inside chat GPT. So to to.

A

um When you're writing this conversational chat, um it uses a same DaVinci model for generating text responses. So let's see what DaVinci would say for us now check this out. It took us almost like a three seconds to get the response, but I like this response. Much better Kong is open source API management platform that provides reliable and Publishing consuming, managing apis provides simple, efficient way to Route secure, so by simply switching different models, making like smarter models and, as you can see more expensive models.

A

So in this case it's a 10 times more expensive that Curie um hopefully like. Let's see if I will be able to call it 10 times like Q, remodel.

A

It's still still stick with the um with this um okay, what is um the best? The best cloud.

A

Native AP Gateway, and uh why it is Quant. Okay, let's see that we, what DaVinci would tell us Kwong is the is the most popular and Powerful Cloud native gateways. Okay, enough said: uh let's see if a Curie model, if I um adjust my request, maybe QD model would be good enough for us see so I was able to um tweak my prompt to get um a slightly interesting, uh slightly different response here.

A

um So we're going to be focusing on this one. Okay, we're gonna focus in on these two models. Since um our developers, we want to expose this API to our developers, and we know that the two models will be different and we need to provide the way how they can consume it and, most importantly, we want to restrict this one way to do. This would be a first of all, let's create a service, so this is the my connect platform.

A

So inside my connect platform, I I selected, North America will switch to Europe afterwards and uh one of the things that we will be dealing with this is our control, plane and inside our control. Plane I will be seeing different runtime instances as our data planes instances of Quant running. So a few few uh data planes instances are running here, uh my the one that I will be using right now deployed latest and greatest version as of today.

A

Maybe tomorrow it will change that runs inside inside kubernetes and and and this this Quant Gateway instance is available for me. If I will go and say that's my URL. If you just hit this URL uh right now, you will get a classic conk response saying that uh response the request not found. So nothing is configured it's running by itself. Nothing is there, since it's a bi-directional communication. I will be able to do configuration in my connect and push this configuration into my um in my data plane. So what I want to start with this?

A

um Create the service, uh create new service version uh version number just V1 and and create it in in runtime group. So this is important, so this service will be configured inside my runtime group that is marked as default. Once again. Let me create this, and I will show you in runtime manager. I do have a default. Runtime group that will consist of two con nodes- one node is 2.8, is very old and it's compatible, but it's still connected. Maybe there's some of the some of the features will not be available, but I'm configuring.

A

This control, uh this data plane and this service will be available, uh will be available here as soon as I will deploy it now. I'm going with this version, 1 and adding implementation, so I will call this open. Ai and URL will be here. We know this URL from this we'll grab it uh click next and I will create the route just the first route, for let's say DaVinci for my routes and the path would be the I mean cheap and I'll create this.

A

uh What was the error.

A

Upstream URL https completion, let's see if, um if I need to do https, okay and I create what is this service? Let's see.

A

A

Will I be able to create this one?

A

Okay, so one more time.

A

So it was able to implementation. Is there so, let's see if the traffic will go go in so if I go to this, this is the address of my Quant. Let's see if we will be able to send um no routes match, meaning that we're doing something I'm doing something wrong. Let's see, let's start over and delete with one and open it up. Let's do start start from scratch. Delete service. Meanwhile, question how the service Hub different from runtime service in Quant connect.

A

So it's a very good question so, essentially inside this runtime you will also have a Gateway services and one of the things that I will be able to do is to have um configure service inside this particular Gateway service. Hub will be used for internal to discover those services and to publishing this in a portal. Let's try this one open AI.

A

This is our URL and, let's see if I'll be able to create the service, okay service was created uh huh now I will be able to create routes. Let's see, I can enable route DaVinci, and in this case this is DaVinci DaVinci route.

A

Okay. Now, if I'll try this, uh let's see what is our our Quan says in terms of like why konk cannot get this configuration I'm going into this um in this config. So let's see it's always always great for me to see if the um runtime instance, if my runtime isn't sync, so it should be in sync- and this route should be configured.

A

Let's send this one more time perfect. Now it works. It took some time to uh to push down this configuration from control plane to data plane. Now it says you need to provide a model parameter so since um uh I can I can do this in put the DaVinci model in every request. I don't want to do this.

A

I want to um provide Simplicity for people, they know if they go and hit this DaVinci endpoint, and only thing that I want them to do is provide a prompt, so I can use this Quant Gateway functionality to enable this. With this I'm go ahead and adding plugin plugin called request Transformer.

A

So what I want you to do every time when we hitting this endpoint, we want to append body to include the model, so in this case I would say, add to body just uh when I submit this to open AI model and the particular text in I'll just copy paste it from my previously prepared from my previously prepared thing.

A

So I'm adding here and if someone decides to Tinker with this uh with this, uh the endpoint I want to also prevent this. So in this case, I will go and replace body. So every time when someone will try to fix the model here, it always will be switching this to DaVinci.

A

Now with this. So when I go and just hit, DaVinci see just a prompt and the max tokens defines the how many you know the the words it will return.

A

And we're waiting, and now we had response so right now. um I'm, including open ID key with my request, but this open ID key will be can be included as a part of Gateway. So you don't have to do this. Let's do the same thing for uh for our query model, so we're doing the query model we're doing with this and we're going with the theory.

A

uh Query model and I enable plugin Also. Let's do a request, request, Transformer, um but the um we're doing query. What was the model uh we go in here. Queries: zero, one and you can find the list of the models inside the um inside open AP, open, open, AI, replace button.

A

So now now, if I will go ahead and key kick this one I need to provide the parameter model um in a few seconds. Okay, so looks like I messed up with the configuration. Let's see if I, um uh with the model, append body and config replace body that should should work. Let's try it one more time, so response actually came in from um from Quang we're doing correct, URL.

A

Okay, that's part request likely you aren't using HTTP Library correct, open, API, spec Json payload. So let's take a look. uh What kind of kind of problem we might have here so I'll just do header.

A

And let's see what kind of Errors we might have here, anything here.

A

It's strange because it uses the same the same configuration that I used, add and replace maybe copy paste errors happens with the best of us.

A

Okay and config replace um replace budget. Yes, correct.

A

I include Bearer I include the application Json, let's stick with DaVinci and I will figure out. What's not what was not working with the um what was not working with the open uh open here? The cool thing is that you know if something is not working. I came here prepared, so this is where I already have my um the the the that configuration file that allows me to quickly um get into the point where everything is working. So let's do this so I'm going into the egg.

A

And what I can do? I can use this deck sync uh and um connect runtime group is going to be default and I will be using my first file. That's called one service route and con, so what it will do it ditched all the bad configuration that I had or might have in in my in my config, so it just showed what it replaced and replaced with the things that I already configured. So it also includes some of the plugins that I pre-configured, so we can go ahead and try try search, try this one.

A

So, let's see if our DaVinci thing is still working, so I can go ahead and just click here, DaVinci model came by.

A

It goes through the clock and our Quantum is open source Gateway now with Curie, let's see um yeah and if I will use something like um what was the what was the request for uh for Quant and for completion. There would be this one.

A

um And also we'll do this one. So, let's see if Curie model will be up and running cool, so um the two things working. So that's fine! So now we do have a two two routes uh inside this Gateway service. We also published this Gateway service inside this open API, open AI.

A

Now what I can do I also go to publish this on a portal and in order to make this even much more cooler, what I can do I just like do upload the spec, so I do have a open, AI, open, API, spec, so I'll upload this here and it will be rendered on my uh portal. So if I go to developer portal, I have a developer portal enabled and right now, I using customization and I use custom URL, so Dev portal.us.com Builders asking registration.

A

So this is where um I have my developer account registered with this. uh With this portal it's called Vector connect, developer and I. Can access to this one and I'll see what kind of services available for me I see this open AI service that that is available. So this is the open AI service that available I will be able to um register my applications soon with this. So the next thing is that um I want to show you that you can configure this stuff with this runtime manager. We go in with default.

A

We're going with the plugins plugins can be also enabled for Global, so I do have a global plugin configured and it was coming from um coming from my configuration file that I had before always have always have a backup plan, because the live demonstration can go wrong.

A

So with this um this, this plugin allows me to put um a lambdas uh or like a lower code inside uh inside my my um my plugin, and execute any particular particular code in this case, while I'm running I'm adding an extra header that shows where this uh response comes from. So if I look the header and says um Upstream host and I know that the it comes, the API call comes from this request. How we can prevent people for doing stupid things we can do.

A

um We can enable um sorry rate limiting on the query um and I can go ahead and say: plugins, add plugin rate limiting and just the regular ones uh we will just do um limit by IP. I can think 60 minutes and size. What is it 60 seconds? Let's do 60 second window, not the not the me not the minute and again, the configuration should be based on Windows size sliding window and based on config seconds. 10 requests per seconds, 10.

A

., it's always uh great to see what is going on with the response. Let's see refresh.

A

um When I click save, let's see what what's the error.

A

At least second minute hour day months, I I have I, have this config in seconds so 10 minutes, where's minutes, say 10. Request per minutes is going to be a validation error. Okay, config seconds.

A

Something is not right with the UI today. Okay, so now, I put this rate limiting on my query and when I send this I should be able to get a rate limiting so have a the rate limiting is working. So the simple request, uh simple request, Transformer, is configured here now. What I wanted to do here is to have a more granular control over DaVinci. Some of the applications that will be using DaVinci will be in production.

A

Some of the development and I want to have a rate limiting for DaVinci endpoint that based on a developer needs. So this is where we application registration comes into play. So when I go into my service, Hub I see my open, API I will be able to enable application registration. Why it is important, so I want to enable my developers who use this inside my um in my site, micro installation, so they will be using this API and also I would like to manage the access to this API.

A

So, if I'm going as a developer in my service, catalog I can find my um my applications. I have a two applications for one for production and for for development. Let me do this: bigger I will be able to to do application registration. So, let's say inside my production, I do have some application that I will be using in production. In this application.

A

I would like to connect to my open AI register for open AI request, access and I will be able to configure plugins here I'm going into my openai where's my routes, my routes say DaVinci route, so DaVinci route I want to slice and dice and provide the access to this one based on user credentials. So I have a test environment in production environment.

A

So inside here, I'm going into my application, say my application test application and also I would like to register this for service.

A

So I will be able to see those things are available here for me now for my plugin I would like to add to one more plugin before that I will go in into my runtime manager and um for default and consumers I will create a two consumer groups. One consumer group, I will call it um kind of gold, so I would provide the the the maximum level of accessing to this API say like a 100 request or like a 200 requests per minute, select consumers.

A

So this is going to be so when I enable application registration. uh It automatically created the consumer for this API. For for me, so this is my uh I guess this is the production one. So, let's, let's confirm this: it has this ID 92.8 and if I go to my production application where's my catalogs, my apps go into production application in 92. It's my production application.

A

So every time when the consumer will come with credentials with this, ID I will be able to override configuration for a rate limiting. So let's save it and I would say Windows size, 60 limit number. So it's going to be like uh what like 20 requests and another consumer group, that I will add just call it silver and in this case I will add this guy.

A

And with this limit configuration we will have five. So every time when I come with developer credential to um DaVinci route, I will be able only get five requests. So in order to configure this I'm going into my go in my test, application generate credential test app um for webinar, so I'm just generate this. One I got the key and for my for my insomnia, what I do in managed environments and go to base environment. So this is my test key that I will be using.

A

And I confirm a copy okay scorpion now in the for the for my route. Davinci now I need to configure rate limiting advanced so rate, limiting Advanced plugin I. Have this configuration called enforce consumer groups, meaning that it will um the find a correlation, gold and silver and based on the key that will be included inside the header? I would be able to route and get different service level agreement.

A

um Just config limit- let's do just for for test window size 60 is going to be fixed now. So this is the moment of truth. Let's see so now, I'm hitting my uh let's make sure, there's now no API key, so I've disabled, this API key so and and I send this request so unauthorized, because API key was not found in the request and now I enable this API key and send invalid credentials.

A

Let's um generate let's, let's test, uh let's test, if the test thing works because I generated for test.

A

So, let's take a look at headers, my I have a request limit of five, meaning that, based on the key Gateway, was able to correlate a this API key to consumer that was registered uh through developer portal and configure this rate limit for developer portal for a developer developer. Key now, if I'm going into my application, let's, let's fix production, generate credential call it prod app for webinar and I click generate copy and configure, um and inside my manage environment based environment in production.

A

And here we will replace.

A

With production key, so what I'm expecting to see 20 uh 20-year request limit, I, replace the production key and yes, so huge success.

A

um Now, last but not least, the one last thing that I want you to quickly talk about is how we can move this into different environments. So now, I'm running this in the North, America and I will quickly switch to um to runtime manager that will be running inside a European union, and my runtime is also deployed there. So right now, there's no Services is deployed here, but I will be able to do this same I. Have this configuration that I already have in my.

A

I can just do deck.

A

Dump so I can a dump configuration that I have in my runtime group connect for us.

A

So this is my configuration that I just um uh requested from us and what I can do I can choose to deck. Sync.

A

um File, it's going to be the same runtime group and now I also need to do.

A

Connect address.

A

So now I just need to specify different a API for our control plane. So I will push this configuration to say deck. Sync.

A

All these things were created in my um in my environment in Europe. So let's see if I will be able to get access to say to DaVinci endpoint.

A

So because I didn't enable application registration, it didn't took into account my rate, limiting that was configured based on consumer group. I can go and do this but rate limiting based on the global route still works, so I think it is. It is a huge, huge progress, so we can. We can create this one, so that was it. It was a lot of things today. So first of all, I introduce you to uh the concept of the um the the connect as a as your like management platform.

A

um I explained you different deployment configurations and it showed you how you can configure external service and pass um uh different body elements. I will modify the request on the fly but including the model numbers I showed you how we can enable application registration and have a developer portal configured and developers can register to service and get the keys and you, based on this key.

A

We can provide different access to the service and, after that, I showed you how I can just simply two configuration from one place and replicate it to another configuration without me to going and doing this manual again so I think it was pretty impressive. With this, um as always, my name is Richard gammov and have a wonderful day.