There is no doubt that distributed tracing is critical for microservices observability. Supporting that claim is monitoring guru, Cindy Sridharan, who recently wrote that “there is no other observability signal that is pregnant with data more rich, causal and contextual than that carried by a trace”. But often practitioners end up with hundreds of thousands of traces and most of them are for normal behavior scenario and hence aren’t interesting. Instead of looking through a trove of normal behavior traces, practitioners want to quickly get to traces that are outliers and figure out why they are outliers. Wavefront Distributed Tracing now enables practitioners to pinpoint outlier traces and understand where they place in the latency percentile scale.
Find and Visualize Outlier Traces with Wavefront
Assume that you’re a developer at an eCommerce company and you’re the service owner of the shopping service. You’re looking to optimize shopping service performance and thus improve customer experience. You look at all the traces involving the shopping service and see some traces which are, say, 5 seconds long. You have no idea if that latency is good or bad because you have no idea where it stands from the median for similar traces. We at Wavefront saw the same dilemma and decided to provide you visibility into not just detailed traces and service maps, but also the ability to sort and highlight the traces based on outlier scale. So, in addition to narrowing down outlier traces via sorting, you can hover over each trace and visualize if it’s a suspect candidate for further examination.
Sort traces by outliers and visualize each trace latency percentile
Furthermore, once you sort the traces by the outlier, you can visualize its service map, RED metrics and complete trace details in the same view. In addition, we often get the request that customers want to search traces with a given tag, for instance the environment or customerID tag. Wavefront provides a very powerful and easy to learn query language with over 120 built-in analytics functions. We already support query-based filtering of traces. Now we have further simplified tag-based trace search, so you don’t have to worry about writing the query. You can just pick the tag from a drop-down menu and visualize traces containing the corresponding tags.
Find Those Outlier Needles in the Haystack Faster
Distributed tracing is an important piece in the puzzle of microservices observability. But sifting through hundreds of thousands of traces is difficult and often not feasible. Furthermore, most of those traces are normal behavior traces, and hence don’t need further analysis. Customers need to quickly get to outlier traces and analyze those further. Wavefront Distributed Tracing now enables customers to pinpoint outlier traces and understand where they stand on the latency percentile scale. Furthermore, it’s even simpler now to filter your trace search and limit it to traces with specific tags. For more information, please refer to the docs and check out our free trial.
Get Started with Wavefront Follow @chhavinij Follow @WavefrontHQ
The post Normal is Boring. Show Me the Outlier Traces! appeared first on Wavefront by VMware.
