What is Tracing & Why You Need It

hi and welcome to this tutorial today we're going to talk about what tracing is and why tracing is essential for your system most applications has regular logs and regular logs are a fundamental tool for us developers to debug and monitor our systems and figure out what's going on as a system grows the significance of maintaining comprehensive logs becomes more and more crucial and is really important for you to retain observability of your system so why are we talking about logs we were going to speak about facing right well it's important to understand where logs will fail us if you have a small system not many moving parts then logs will probably be just fine but as your system grows maybe you have multiple services and many microservices which are triggering events which triggers other services to do things so how will you know why your system is behaving in a way that it is for instance consider the error logs that we are viewing in front of us three different Services we have a register an email and an invoice service we can see that a new user has registered these are just plain logs and we can see that it sent user email about being a new user it did send a user email from the email service but then the invoice service is sending an email and all of a sudden we have an error in a log system this error would probably be picked up and we would get a notification saying that the email service has failed to parse the template maybe in a real world scenario your log would have a little bit more information but it doesn't tell the full story and that's where tracing comes in if we only see this small error log saying that an email failed to pass it doesn't really tell us why our system was trying to send an email or where the email originates from in smaller systems this is of course pretty trivial to figure out because maybe you have one or two functions that trigger emails but in a large environment it might be really Troublesome to understand why the template is failing tra in would allow us to visualize a flow of operations throughout our system one thing that is great about tracing is that it won't only show us that an error occurred it will show us why it will show us which service triggered the email service to actually send an error the logs that we were watching is actually from this scenario we have a user who can register on a website and our register service will send out a new customer event to the email service which is also consumed by an invoice service which will then send out monthly invoices in our scenario we are only seeing that the email service fails but we're not seeing why and we can't understand how so we will take a look at how we can Implement tracing and how tracing works so we can figure out this scenario so tracing is really actually in practice quite simple imagine if we were to change this into this every time we sent or triggered some operation we would add a unique tracing ID and this tracing ID will flow through all our services we can use that ID to tie a workflow or a flow throughout our services so that actually this ID is sent between Services allowing us to step back and Trace why this event is happening so as you can see here the user registers we have a tracing ID of 1 to three this tracing ID keeps following along our application to the email service and the same thing goes here to to the invoice service we can see the same tracing a month later when an invoice is sent that's a new operation that's a new trigger for a function to be triggered it creates a new Trace ID as we can see followed by these arrow down so a trace ID is pretty simple it's a unique identifier assigned to a single transaction that spans multiple services and the trace ID is used to tie a full flow together from start to stop between all the services imagine a bus the bus is going from point A to point D and it will trigger a bunch of events uh and stops uh in between so the trace ID is basically the route that the bus will go so one other thing that we have to cover understand tracing is something called spans spans are basic units of work inside of a tracing flow so imagine you have the bus the bus route is the trace but it makes stops along the way to do certain things let off passengers switch driver Etc all these are spans so if we take a look at our register service for instance our register service will update the database and that update might be a few operations maybe an insert to the SQL database or whatever you're using so that's a span it's a certain task at first you might think why make it so complicated but it's amazing because each span we can tie to metadata start and end times names information information Etc and this is really really great because it allows us to see bottlenecks in our application what is taking time why is it taking time and one thing to to know is spans allow us to add as I said metadata spans can also contain shy spans so these can be nested so a span can contain a few other spans inside of it so both Trace ID and spans are very essential things to understand when you're going into tracing and basically the goal of tracing is to allow us to build a visual map over our system and understand which events triggers what and the flow of the data inside our system open Telemetry is a set of tools set of apis sdks ways of doing tracing combined and gathered it's a way to unify How We Do tracing and how it should be done most of the systems does support open Telemetry and it's just a great standard to follow in the examples you're about to see I will be using Jagger to visualize my tracing let's go back to my terminal and I'm going to start up my services now you can see Jagger spitting out a bunch of information I'm I'm also using knots to propagate events between my services uh anyway let's go ahead I'm going to send a curl request to SL register so if we go back to my logs we can see the logs printing that a new user has registered and then we have the error the HTML template is not part let's just go ahead here and search for all the traces hi everyone I just wanted to thank better stack for sponsoring this video If you're looking for topnotch log manager ment and monitoring better stack has got you covered better stack offers powerful tools that help you monitor your entire infrastructure analyze logs and ensure that your applications run smoothly better Stacks intuitive dashboards and realtime alerts make it easy to stay on top of any issues before they become a problem beack can help you monitor up times help you with monitoring logs and metrics not only that integrating better stack into your infrastructure is really easy by by using their interactive connection setup you select your platform and there is a ton of platforms supported everything from De to Docker rnet AWS and a lot of other stuff basically everything is available so if you need reliable uptime monitoring and log management do like many of the other big companies and check out better stack thank you for sponsoring and we can see quite clearly here that I have have here is one trace and here is another Trace you can see the trace ID up here so B7 o is my Trace you can see the first one contains five spans so the whole trace route had five stops which did something we can actually see all the services which are involved inside this Trace so we have opened up our first Trace here and we can see that it's originating from the register service and whenever we did that it started a bunch of sequences of events inside our system first we got the information to/ register then it started to insert the user information in the database I can open up this span and see more information about it you can add information to spans and metadata whatever you want it all depends up to you so we can see whenever I do update invoice we can see the invoice ID we can see which operation it recalculated the invoice uh and we can see the user ID this is all because I added data to the span we can also see that we had one event triggering which was save a invoice so each span can have events which is sort of like tasks that happens inside of the span so it becomes pretty clear what's going on inside our system when a user registers and then we can go back and look at this one we can actually see we have one error occurring inside this Trace but we can actually see the other related service is the invoice service so we won't have to figure out why the email is failing because now we know that the emails are failing when they come from the invoice service just by looking at this we have Tracked Down the origin of the bug but I hope this shows you how amazing tracing is really I don't know how many times I've heard people say like yeah but this error only occurs sometimes and we don't know why every time you hear that you know that tracing would probably have helped you I do hope you enjoyed this video and I do hope you understand what tracing is and how can help you and I hope you have a great time implementing tracing take a look at open Telemetry there's a bunch of help out there to get you started in whatever language you want to go

This video is a conceptual video that explains what Tracing is and why you can benefit from it. In the video, we don't go into the details of how to implement tracing, but we look at tracing and learn how it works and the idea behind it. In the video we will learn more about what Open Telemetry is and how it works under the hood. We will learn more about what Traces and Spans are. A video on how to use Tracing with Open Telemetry (OTEL) will come very shortly! This video is sponsored by BetterStack, a monitoring solution for almost any infrastructure or application. Look at their website if you want to get a full view of your logs! https://betterstack.com/