OpenTelemetry demo app with Grafana, Loki, Prometheus, Tempo (Grafana Office Hours #06)

hi everyone and welcome to another grafana office hours I'm Nicole vonderhoven and I've got a co-worker with you today that you probably haven't seen yet at least on this channel oh I've been on here I think so sure I'm Paul balog another one the developer Advocates here at grafana labs and this is my second time I think on the grafana office hours thank you very much Nicole good you know exactly memorable sorry about that but what is memorable I mean the shade is starting already but what is memorable is that we have a special guest today and his name is let me try and and pronounce this when Liu he said that that's how we should we should just call him when I'm sorry that my Mandarin isn't up to scratch welcome to the stream one yeah oh okay hi uh hi Nicole and Paul and everyone my name is my Chinese name is Leo and I'm come from Taiwan since my name is not easy to pronounce will be fine and uh currently I'm a Deva engineer for about three years experience in an insurance company which is the biggest in Taiwan and we have has more than a million customers and my memory is including maintenance crcd a python City Pipeline with Jenkins and the kubernetes and they also introduced the tools to enhance our developer experience and the first cycle of the devil's Loop beside the work I also contribute to some open source project and write some package for Mk tax which is a static website generator one of my parking just got 1000 user recently yeah oh that's awesome thanks nice like how how did you go on Paul oh no I was just going to ask so it's a static site generator like uh Hugo uh what what language was it in did this uh yeah yeah yeah it's like you Hugo and uh with the MK ducts we use so you write markdown and you generate the website with python it's a python package okay cool yeah you can do a full site and publish our GitHub or or anything else everywhere anywhere else yes for well I I actually didn't realize that it didn't like click with me I might talk to you about that offline because I do other work in in markdown with another tool called obsidian but anyway that's a different Channel um how it started with grafana in the first place like there's a lot of there's a lot of visualization tools out there um when was the first time that you started to use grafana the first time I met grafana is maybe about four years ago and that time I was a better engineer a responsible for building a machine learning API service with python and then introduced the python API framework to my company after the service was built I want to know what the performance and the resources usage because when you use the machine learning service the the performance is really important the user want to the responses to slow and then you and I have to check the CPU and memory it's now over the limit so I started to look for uh to tool to monitoring those metrics of my service sound research I think um when you find something one to uh monitor yourself is the the first half hour will be the promises and uh gravannah so like everyone I I found a girlfriend and the promises so I started use land and uh I think this is the first time I I use Google Now start with the promises and the loss Matrix dashboard yeah and that was actually how I I was connected with you because you entered the the golden grot contest and this was a contest that we had for dashboards grafana dashboards from the community and wet here was one of the finalists can you talk a bit about what dashboard you made was it this one that you submitted the one you're going to show us okay so let me show you first so so was this the one that you wanted to was this what you submitted yes yes yes so you can pretend my my screen okay sure you can switch to it okay so this is my notes for me to the the the contest is the it's the dashboard if not have too much too much uh uh this time because the the basic idea is I want to show how the grafana can can show The Matrix and the logs and some trust information like here so I build this dashboard and I submit it because I I think maybe it's just a try nah nah I have no don't have too much hope to on the finalists or you can custom price but it's it really cool I I color on the final list yeah can't sell yourself short you know what would have pushed you over the edge probably too is if you had some flame grafts on there because those are showy uh everybody likes to see flame graphs you would have probably won yeah so maybe so this is already like part of the the app that you set up so I guess before we get into that why don't you tell us like why did you want to create demo apps in the first place was this like for learning or was there a specific purpose for it yeah okay so about two years ago ago I saw a Gruffalo webinar introduce how to use uh mattress and the logs at the same time with the Explorer feature you can use that hook query to find the Matrix and the narrow down to the time range you want to invest the cat then query or send time range with the sync feature in export app to find query the logs so uh I was impressed by the palace creature because in my company we we stole our logs in a bunch of files and the separate life uh by the file size it's really hard to find a lot you want to check for example when if you want to find some time zone or time range uh when the record request uh have to have much to have a lot of requests so you have to combine the metrics you have to find the time time zone from The Matrix Matrix and then then from the front time zone to find to go to the log file to find which actually time zone the logins but with the the feature with the thing feature you can you can pretty easy to to find a log on the on the sensing side on the profana so I think this is can help me a lot then I started to search how to how to do it after some researcher I I started to know there is an idea called accessibility it means you can you have much more more information to know what's the going on on your app and the Covenant provide a grad great ability to to to see this this information on the same price and the and then they provide a triangle with the metrics logs and the transits so you can query them at the same time and make land have the more power when you can only query you only one information talk about the triangle in the post they also provide a demo project called anfx but the the demo project is written by written with uh golet but my use case is uh prices and I'm much more familiar with python so I studied to build my own demo project with python and I want to make the demon project as simple as possible in the red and they're ready to use Alpha box so I use Taco compose to Orchestra for in front like confrana Prometheus Loki Temple and my demo python API service I want to use this demo project to show how how how's the what's the ideas of the stability and what they can provide some strong ability to to you have a better understanding of your service so okay so I what I like is you you so you mentioned some issues with setting up an observability stack because I think um sometimes it can be a little bit daunting when you when you want to to make your application observable it's like where do you start you know one of the things is how do you get data from different components the modern architectures these days can be quite complex and so to to determine like how to actually instrument everything um that's one of the things and then you also mentioned the problem of how to correlate different types of data because every component is going to have like different types and maybe have different requirements and then how do you kind of match the two so that you know like what was happening on every component at the same time yeah I think uh the the I think the most difficult thing when you study what's the absolute abilities is it's how to collect those data to maybe a single platform I mean you could find out or other platform uh for me the first thing another easy is the metrics because almost every every uh framework or language you have the Prometheus kind so you can expose those metrics to uh of your service so the first thing will be the The Matrix then the second thing is the logs so I I use the low key and to to collect to collect the no to store the node notice to store account and using the uh Lucky Duck driver to collect log from from the container and then install then collect the look and pass it to to the local so I think the the most difficult is the chest stress the information is much more difficult when you when you want to add ability to your to your monitoring systems uh I think uh because and and also using kubernetes I think maybe there is a lot of people who are using kubernetes they were more flavor they maybe have the maybe they maybe have used the istio now you still have the actors and demo project is called the book info you can you can install it and you can see all my info my there is my a lot of my service python service a real service or no service then you can use the istio to find on the the uh the request is sent from the service to lotteries and then to this service that is uh address the hospital addresses but the the but maybe better I think the the demo project will make list is too too easy things like magic you you just you just you just have a CEO of istio that you can have the full chat but when you have the but if you want to have more detail actually you have to maybe uh do some do some install some library to your application and uh edit your call to have more information about the chat so I think that is more difficult the difficult part to to deal with the trust information yeah yeah I mean it's like uh yeah with istio and then other like Linker D and all that uh those service meshes they yeah they definitely give you that not dependency graph but a relation graph where all the services you can see that what calls what uh things like that but yeah it's not usually enough information you need you need more information about some of the requests that's where the tracing back a second and and Define sorry can we Define what what a service mesh is pull hahaha yeah no um wow yeah service smash how do you really describe that I I almost have to go to chat GPT or something real quick but uh no I mean but uh but yeah but I I guess uh let me let me explain it the way I think of it is you know with the service mesh you have these agents uh essentially they're are they're just processes you know running in uh alongside you know as a side card usually in kubernetes so you'll have your pod and then you'll have an extra container that's running this the service match the the agent and what's that what that is actually doing is it can proxy requests and so that uh you know if you have reliability issues or things like that where a downstream service is maybe not available or or the one particular instance you can have it switch so it just kind of I don't know it's like a watcher it's a it's a separate process that sits there and then it watches your application that's in that pod and and gives you uh reports for you some of the information it can you know forward logs it can provide that some of that Trace information uh things like that and they can you know like I said it can do circuit breaking and yeah basically you can do all kinds of stuff so yeah that's that's the real short answer would you say that thanks for putting me on this and I'm asking for myself as well well I'm here to ask the the hard questions so would you say that it is a form of instrumentation I would say so yes yes I mean well okay because it seems to me that it's less or passive I mean because the the application author you know for what's being watched doesn't have to do anything with it when you have the service match they don't have to do anything extra they just write their application they may not even know that there is a mesh involved you know so it is a little bit more passive in that regard but uh but yeah but it does provide you can provide that instrumentation with it it can sit there and monitor the health you know the health of your application and say that yes it's it's all good you know so that's a form of instrumentation in a way uh uh you know forwarding logs and things like that I don't know did that answer anything more questions yeah yeah makes sense so I think that there are mainly two approaches to monitoring and I would say that one is instrumentation and another one might be ebpf and um why don't you talk about instrumentation is you don't have to you don't have to to edit your code you just use then like uh in Java they have the agent so you can just uh up the agent to your application and you don't have to edit your code so uh and then the people have as far as I know uh they are more close to the Machine level so you have so yeah so they're monitoring how's the going on on the bed on the OS label [Laughter] um yeah no I and yeah I know like like this much of ebpf unfortunately um I know how to spell it uh but uh yeah no uh I just know that again it's a little bit more passive to the application I mean because it's it's monitoring on behalf of that application you don't have to actually put code instrumentation code in your app to get the benefits from ebpf so here's here's the way that I think of it and either a few or anyone in the comments can tell me if I'm wrong sometimes it's easier to put something out there here's how I think of it I think that instrumentation is usually so imagine you have different components instrumentation usually involves having an agent like like Paul said on each of those components and then exporters on those agents that allow it to effectively speak the language that that is needed to be able to send those requests out to some Central service that has advantages and disadvantages one of the main disadvantages I think is it's very specific to protocols or to that particular format right um so there are many many many exporters and it can be difficult to find the right one or or sometimes there are versions of the same one and you have to have exporters for each one the other approach so this is more like I think I think that it is um closer to its own it's closer to the application um ebpf is on a way lower level it's it's closer to the kernel so you're it's kind of like you're instrumenting the kernel to some degree so instead of sitting on on um the application Level you're sitting way lower and instead of of having to specify having to have exporters for every little thing you're instead the whatever thing whatever tool you're using is just going to be able to do it based on some standards like maybe it's all HTTP traffic or or whatever it is and it doesn't really matter what is generating that traffic it's just listening in general so it's kind of like you know if if you were if it were um if we were humans I kind of think of you BPF as going down to the molecular level and seeing what's happening at that level now you may not know like what the person looks like from the outside or you know what their names are but you're going down to their to the very core of them that's kind of how I think about it now what part of that was wrong laughs was that was that did that sound right it sounded great to me um yeah um billions and billions of tiny molecules uh [Laughter] so let's also talk about let's dig into instrumentation because that's the route that you chose when for your demo apps um I think it's impossible not to talk about open Telemetry can you can you talk about open Telemetry what is it and why does it matter is uh uh to upset and include the API and the SDK and some collector um I think the moment in which is maybe about maybe a few years ago maybe people know much know much better known Yeager and maybe on the 15th I think can you and uh but uh after some after these years people are moving to the open television and provide a better uh SDK and the unified the protocol to get people to commute to to passing the traffic information they also have deal with the logs and the merchants but the memory is the traffic so I choose to use the opening Matrix because the the library the SDK is is so many people are using using it so if you have some [Music] whatever issue or some government issue there are a lot of people can can do they can can fix it for you so I choose the open Telemetry and when uh open to immunity profile a lot of I have two two ways to to ensure instrument your app application the first one is the auto auto instrumentation you don't have to to edit your code you just use the the use the the make mechanism provide by the language like Java Java have the agent agent feature so you can just just cook the the Java agent so your application can be instrumental and the other one is the manual instrumentation and you have to you can if you have some respect uh some specific usage thing so you can up you can use the the manual instrumentation to collect the data or or update the data if you want to collect those instruments to to collect those stress information yeah I don't that autonomous one is uh pretty wild in the Java land I know uh with some of that spring boot magic where it's yeah oh wait I just have this jar on the class path and all of a sudden I'm getting all this extra functionality yeah it's pretty mind-blowing sometimes I like the explicitness usually so I think that open Telemetry is an attempt to kind of soften the disadvantage of instrumentation that I said before the the fragmentation of it the fact that there's so many different formats um and exporters that you need and at some point someone was like hey how about we just all agree to speak the same language and say it in this syntax and that way we if we want to make it a make our applications observable then we just have to make sure that they listen in on that frequency or or that they re they use that same syntax as well and they don't then the people like the devops engineers who are watching our applications don't have to go and and make sure that every single one is um is going to be able to be observable basically um so I think that's a really important point that you know Open Standards are really necessary because open source is so like it's there's so many there's so many advantages to it right but one of the disadvantages this is is the fragmentation of it because anyone can create anything and so you end up with so many different ways to do things and you can't agree it is important to agree though because it's only when we agree that that we can start to create things that are interoperable you know it's like like soon we need to all speak uh Esperanto so we're all we're all speaking the same language yeah the American that doesn't know no album I'm sorry I'm just picking on top in particular it's normal [Applause] [Laughter] um so tell me about yeah go on one yeah for me I think the standard uh the standard is protocol is really important because uh for me I'm a devil engineer so I'm not really good to develop the application and we have a lot of language in my company and I only know my my much more familiar only familiar with so when other things then use other language we I have to convince them to use the the syncope standards to to to collect this information so I have to find uh we have we have to to just just one to to collect the information from all of my company so choose the the most people use the center and then I have a much of the Community use so provide a lot of SDK and the example and the faster issue issue has has better issues dealing speed is very important when you choose the standard yeah maybe we can jump into the fast API app and maybe you can tell us what what it is yeah okay well let me show okay yep we can see your screen yeah yeah okay so yes this is one of my demo projects it's called Uh take care accessibility the mango is one to uh demonstrate how how do you enable the accessibility to your first API application uh API is a framework of a python so the the the the the idea is it's pretty easier you just have your one this is your vacation so you have a lot of you have three information to want to observe so the first is the chest this one the application a little bit by the way I think nickel knows just in case going closer yes I was a good intervention any names but yes much better so the we have the lubrication so we have the the three period of maybe when you when you study lots of beauty they have talked about the three pillars of the field they they are dressed and the Matrix and blocks so we collect the traps with the tempo so the application will use the Telemetry SDK to push the traffic information with the grpc to the temporal service and the second is the Matrix Matrix we everyone almost everyone is and uh in Python we have the communities right with the with this Tyson crime you can enforce uh endpoint call Matrix slash Matrix then then you can set the Prometheus to describe the Matrix from your application okay so the permissions will responsible for the capital Matrix indexer is the look s so we we are using the Noki because availability to to query the nodes from Okay so lucky is a responsible for to install the log and there is another plugin called darker local parking to collect local from the our application okay the application is from here online container so we are using the Tactical concave so when you collect all these free information you can query from crazy information from Okay so eventually you will got this dashboards so you can see oh there is some metrics please know this is the total request and the this the duration of this API and uh his downloads and then you can here you have the test information like this you can see the address information is correct to the tempo and then you can query and uh on network funnel yeah that's cool can you um Paul go ahead no I was just gonna say that traces are always impressive on demos and that I mean that's that's always the coolest thing because it's like to be able to dig into that level and see all that information you know the path that a request is taking is uh it's yeah useful it's very impressive always yeah so maybe it's a good good time to talk about like the differences between those those three the triangle that you mentioned logs metrics and traces I really like tracing because it's a way of being able to um kind of tag a certain request and see the path that it takes throughout the application so it's kind of like you know a car that has one of those E-Pass that's e-passes that go through the tolls and it goes beep okay that leg took this amount of time you know and I mean it's not so good when they're charging you for it but it is good when you're a devops engineer wanting to know which part of that entire Journey took the most time what about what about metrics in general what what are metrics I think I think my message is uh you can in my opinion I think I think we act like like the heartbeat of my service so I can see what's going on on my service for example you can see how many requests into your service and their requests right and the status uh what's going on on the surface maybe to have some issue like this this excuse to Too Much Time about five five seconds to to to get response so I think that it's like a heartbeat you can you can you can actually see the the services like this live or maybe maybe it's done you can see this when everything is counter zero yeah I kind of think of it as like the dashboard on a car you know this like the fuel gauge and and the speed and and things like that so the things that you know that you want to track all throughout the journey but if you only look at that then you don't know where the entire car went you know so you might if you only look at metrics then you might see for example oh the car was going really fast here but where it was going really fast or really slow matters you know because maybe that car was going too fast in in a in a school zone or something or maybe it was going too slow for a highway but it wasn't going too slow if it was in a residential area so you kind of need to know where that was where it went the tracing part in order for the metrics to make sense because a metric on its own you know is is can be ambiguous so what about the logs component so no uh I think I think is the best best thing of our vacation because before we don't have a mattress and address before maybe 20 to 20 or 30 years ago the notes and it's the most important how how do you know how to know the application is going because you can write a so I think in in this case when we use an opportunity we can the the power of notice we can use uh query the logs when uh at that time with the Matrix hello no no I think we're all good yeah okay so I think I think it is uh inaptability we can we can we can we have the we can get more power more power to our loads for example you can query or if you want to want to navigate this time rate you file you will find there is to process around of the service and you can select this range and then you can find this time range laws so I think this is the use case use the Matrix and uh the track and the logs together like I mentioned people's uh I saw I saw the feature display on uh yeah and then the other one is uh you can you can combine using the notes and the address for example when you you can you can add the trash chat ID to your to your locator so you can you can know this log is is is produced because of this this request i o Tesco so when you when you have some is too slow then you can find the trace ID to find what is the the log base so I think this is a better way to to investigate what what's going on yeah and I like how the uh you kind of went over real quick but it's it's kind of a big deal and that like you had the uh the log there and then you were able to click on that button which drilled directly down into the set of traces based on that Trace ID you know so it's kind of nice so you can you know to be able to go back and forth and I I thought I saw it there also too on the uh the trace view where you can see the logs there was a log that yeah there it is yeah logs for the span so then you can uh you know go the other way as well I don't know the tracing stuff always it just always impresses me how well it demos because yeah back you know the the 20 30 years ago yeah I was there um I was waiting for a comment about uh some old guy uh but uh you know wow anymore I know I know it's you know we had our logs and we loved it it was all we had you know but uh and you could do things where you could like you know create create metrics off of those logs by you know having something go through there and you know do counts or you know summarizing things and create your own metrics that way but but now with the open Telemetry stuff hooking that in there you're getting the Metro extreme you know directly and you're getting all the traces it's yeah it's a wonderful time to be a developer so I was wondering like in the car analogy if if tracing is like you know the the past that determines where it where the car goes when it passes through certain tolls and metrics are are the things on like the speedometer or on the dashboard basically what would be a good thing for like a good metaphor in that analogy for um for the logs maybe if the car were like able to say maybe it was like if it could say oh that was a bump or something or okay now now we hit the break here or now we hit the gas it would be kind of like if the car were able to talk yeah yeah I think that I'd be [Laughter] so can you tell us about uh tracing like you're you're talking about the spans we can see there logs for this band what is a span when so uh the thing is so Spain is this is her husband let me plan okay okay so we have the a lot of the the we have a lot of can see is one two three levels so it should say it can come can become combined to a full chat so your chest maybe have a lot of fat so let me find another one this is this one so I guess a span is kind of like the distance you know the one leg of the journey so to speak so you can you have a lot of uh like this now why is it important to look at Trace uh all the data that tracing gives us through spans so um again um I was just asking why it's important to look at the tracing details through the spans like what can we learn from the spans why is it separated and disbands oh okay I got it so for police for this Trends we have really crossed a lot of service like uh atpa activity so we can analyze so you can see this this space is take about one second to to deal with this process so with this spring information you can you can find uh this one is to take too long uh uh take too long to on the for process for the trend service and besides you can you can see what's going on you they have a lot of information about of this this and like the status or something else I decided uh there is thankful it's fine at road test so what are you doing here so for here for example if you have some some uh you can you can you can find some Arrow arrow on the on the test I understand so you can find is some error information on the understand like the what's going on on your Apple application like here we have a some variable yes so you can see the address so I think this is the ability of the spend information okay or some detail like the query is the service come from The Host ip.com effective so what we're looking at here it's it's like it's um I'm going to I'm going to keep building on this car metaphor by the way it's kind of like this is saying hey somewhere between here and here these two tollways the car broke down and if that's all that you knew that wouldn't I mean that would be somewhat useful because at least you know where to look but I think you really need the the other stuff like the logs what was happening at the time and metrics like did you run out of gas or or something like that to kind of give it a little bit more context yeah yeah so he'll actually we still we also have uh this one knows to outputs maybe we don't have dialogues so like here we can we can drill down to the monologues about these Networks if we have some errors on our lock systems so what did you have to do to get all of this running how did you actually instrument everything um so as previous as I say I want to list uh Emma project is a single as possible so I use the type of compose to to extract all the service so just you just have to use Docker you can have a lot of list service like like lucky and Prometheus and the tempo in the group Runner the structure is just a basically it's just a look like this this picture so it's really easy to use maybe the difficulties the first thing is you have to install the Logitech driver so we can connect the local from our container and just uh like everyone use Docker you just can prostitute and then Target the compost up then wait one second you can have enough for the the four dashboard monitoring the statement project yeah and then like uh with the way that you instrumented the code um using that uh the python client for the open Telemetry data um what did that look like in your in your code um you know how how intrusive was that client I mean is it is it looking like you're just writing an Hotel application or something or how you know how does how did that affect your code in the application I think uh in Python we have to we have to use some uh previous I said uh we use open Terminator SDK in the land that two ways to to instrument your your application we use the manual manual instrumentation so we have to actually we have to update our core so here we have to uh setting up some opportunity uh the service so we have to tell you which Tech Telltale add some information to your spend current information like you're a service man your company service name and then we have to tell the SDK uh where is the endpoint you have to where the location you have to send the address information so we have to add this this piece of code to our application yeah so at least we can we can process the this first step to to pass the the trend information and the Matrix is much easier because uh python Prometheus Tyson Kai have a lot of do those work for us it's really easy right so with so with the tracing you had to be pretty manual so you had to create that uh you had to use the SDK to to I guess uh initialize a provider um and then you had to explicitly make calls to say that hey this span is starting or or you know if you have sub spans I should say um because a lot of that the it'll create by itself right and then here you were creating like uh you were adding those attributes on there or the service name and the compose service and that and those become I guess uh I think it's still is it still called baggage I I think uh back with the open Zipkin I think they always refer to that context information as baggage um you know but uh anyway but yeah but that's where you can actually add more attributes to each one of your spans right yes yeah I I uh actually you don't have to use the you don't have to to tell uh to to where is the space starting because in fast in fast API they have the mechanism called middleware so you can you can you can step the open Geometry as you get to a middleware so every every request into your service and how to out and go back to the user so the spanware of the auto credit um but you you can create manually spans for if you want an even fewer brand or something you know like I don't know for some reason you would want to measure and see how long does this Loop take uh for some reason you know you could create a span around that so okay I did see something though and I understand because of python why you're using it but I saw that you were using Locust have you tried k6 that was yeah that was a trigger yeah foreign [Music] I think it is pretty cool because you can Define the the whole the whole uh the test test prayer with the code and uh and then it's easy to use we're gonna hold you too because I mean everything else is grafana stack so you need to use the uh the reliability testing uh tool from grafana as well yeah well I'll tell you something that you you might like when I don't know if you knew but with k6 there's actually a um away now so that you can instrument the HTTP requests and it works with yeah so you can start the trace from there from the the load test itself yeah yeah tempted to use it now [Music] [Laughter] so we're talking about the the pillars that you're you have put on here Trey says logs and metrics but there are actually other things that we might want to know right like is there anything that you know if you had enough time or whatever that you would like to add to this to kind of round out this educational stack so I think uh the I think the demo project is like a home lab level so it's not really scalability because uh in production maybe you don't you you definitely not using The Deco that are composed to run your run your assistance so kubernetes [Music] so I think we at least with these demo projects we can know how to how to pass this this how to pass this data I think the most important thing is you need to know what the ability how how is a possibility to the data to collect like you you like the trust data is is provided by the vocation or the or the side part of your from your agent and then the Matrix is collected by Prometheus they are collect the metrics from the application and the load loads is maybe collect by other tools like here is the docker logic parking or maybe you are using the front bid or the frontal so I think uh with this demo project you have you much know clearly uh what's going on the path of the data and when you want to build a more scalability on the kubernetes that you can use use those clearly service on kubernetes and dual language this data flow you can then you can have much more scalability uh observability system for your for your for for application so we have a question here from Enrique levera who says when did you use the open Telemetry collector to send all to grafana oh I have other similar project I really love to do this to this demo project can let me know much better about what the needs to is working so here is another one demo project so it's uh it's for jaeger but uh we're using but here is this is the open Telemetry collector so you can use the open image character to to collect Instagram it's like a profit or a engine you your applications and chess information to the opening room to collect then uh collector can process those information to back a chest back and for example here is a year girl or you can you can uh sending push it to push the trend information to Tempo that is okay because open University support a lot of format so when you uh when you collect those trading information in your backend then you can query those uh queries or Trace information from grafana or from query query from the with the profound so why did you use so it seemed like in in the previous one in fast API you didn't use the hotel collector right instead you manually instrumented it yeah because uh uh I want to make men so I think and to add too many components will make will will let us no can can really understand what's going on yeah you have a collector Temple you have a robot you have a lot of things and so I think I want to make a simple can you talk a little bit about your other your other project um the other one is the the same price no set have the same idea whether with the Supreme Court application so you can see it's nothing different the only difference is the application is different but with this input as previously we said uh continuation by ultimate automatic instrumentation so with the ability we can we don't have to uh to rely the spring boot magic yes just you just you to you to run it with a Javas then you're from your your spring Brew application can have the full ability to check your application yeah I think it is pretty cool and uh because my my my most of my colleagues they are using using Java at the right side the same blue approach swimming pool so I I also created this demo project for land okay so they can know what this is pretty easy to to make our applications also available so we are we are coming up on the hour here but I did just want to ask um if you could talk to yourself three years ago you said you know someone who's just starting in this role what's just one bit of advice that you would give them one bit of advice I think uh it'll be go to Cortana the best idea is wow Gallery I think that is really impressed because we can see a lot of ideas no matter how you can monitor the the best or application is this grafana play that you're talking about uh no no no no no I guess that was just the community yeah yeah go to the gallery to find to find the how people to to monitoring their s yeah I think the headline is well I actually didn't expect that that was such a plug thank you thanks for for joining us today when um I I was excited to be able to connect with you after you became the finalist and um thank you for creating these awesome educational apps so that other people can follow your footsteps definitely it was awesome having you all right well we're going to say goodbye here and have a good weekend thank you everybody and we'll see you next week bye-bye

Channel: Grafana

Views: 11,277

Keywords:

Id: dXR8WNm5uos

Length: 60min 47sec (3647 seconds)

Published: Sat Jul 29 2023