Traces: Performing a Comparative Analysis with Dashboards

In this video we are going to build a dashboard that will help us perform comparative analysis against traces. So let's go to dashboards and click on traces demo. This is a folder I have created and I will add a new panel. I'll name this panel as “demo traces” and the stream type will be traces, and the stream will be default. For the y-axis I want to perform this against duration, so I'll select duration. In OpenObserve the duration is usually measured in microseconds, so we will change the unit to microseconds. Then I will remove the decimals and click on “apply.” Now, to get a better visual understanding, I will change it to a line graph. For the legends, to understand the metadata for what traces you're trying to compare, I'll select it to be on the right-hand side. I will keep the width size as 250 so that we have enough space to read through the metadata here, and then I'll scroll down. Now the goal is to add comparison against the last 2 hours and 3 hours. We want to compare the latest with 2 hours and 3 hours old data. I'll quickly add 2 hours, and for the next comparison I will probably add, let's say 6 hours instead. Now I'll quickly save this, apply this, and you can see that I have the latest, 2 hours ago, and 6 hours ago. You can also see there is a drop for the two hours ago, and for the duration it is high right now—between 10:30 and 10:32 it is high. If you want to get more visual colors here you can scroll down and change the color palette to “palette classic” and then click on apply. Now I will save this. If you want to have proper analysis for this one, the highest we are comparing against is for 6 hours, so I will go and increase this beyond 6 hours, that is 8 hours in this case, and click on apply. As you can see, there is not much difference here, but there is a slight difference for the latest. For 6 hours ago, the color coding actually helps—it is a little bit aligned with the two hours ago. So this is how you perform comparative analysis. Now what if you want to compare this for a specific service? For that we are going to click on the dashboard settings, add a dynamic variable, and add a variable—let's say “service name.” The stream type will be traces, the stream is default, and the field will be service name because this is the service name for which the traces are ingested in OpenObserve. So I'll click on save. Now I have to edit my panel. Over here I need to add the service name as the filter. So whenever we want to select a specific service, that will query all the traces for that. I will go to conditions and say the service name is equal to the service name we added in the dynamic variable. Now after saving it, I can compare this trace timeline for multiple services—for ingestor or even for the query. If I click on refresh you will see the trace timeline. Now this has a very good comparison for querying. The current is completely high, whereas if we compare with two hours ago and six hours ago the ingestion was completely low. So this is the best way you can perform comparative analysis.