jaegertracing/jaeger

[Feature]: Augmenting Trace Timeline visualization with Aggregate System Statistics for faster Root Cause Analysis

Open

#6814 opened on Mar 5, 2025

View on GitHub
 (11 comments) (0 reactions) (0 assignees)Go (18,974 stars) (2,326 forks)batch import
changelog:new-featureenhancementhelp wantedui

Description

Requirement

As a user of Jaeger, given a trace that violates the SLO (i.e. 95th percentile latency), I would like to quickly narrow down the spans/operations that are likely sources of the violation.

Problem

Currently, there are 2 issues that prevent Jaeger from accomplishing this task:

  1. There is no visual cue for the user in the Trace Timeline view to figure out which operations or spans could be the cause of an SLO violation.
  2. Jaeger does not currently calculate the relevant data about each span's latency in the context of its overall latency distribution.

Both of these problems increase the manual effort required for developers when they are doing Root Cause Analysis on SLO-violating cases.

Proposal

Inspired by TraVista, @stolet and I propose augmenting the Trace Timeline with span-level aggregate behavior.

First, we propose augmenting the Service & Operation pane to also include the percentile ranking for each operation right next to the operation name of the span. Percentile rankings that are higher than a user-defined threshold (e.g. 95) would be highlighted in red to visually indicate the anomalous nature of the request. Here is a mockup of what a percentile-enabled pane could look like:

Image

Second, on clicking the anomalous span, a new latency distribution graph for that operation will show up so that users can easily see the latency distribution for that operation and where the operation's latency lies within the context of the entire distribution. The graph is a histogram graph with a log-scale y-axis that measures the number of requests that are in that specific latency bin. The bin containing the operation's latency is highlighted to provide a better visual cue to the user. Here is a mockup of what the latency distribution visualization could look like:

Image

To support the above visualizations, the traces collected by Jaeger would need to be further processed to extract the necessary statistics for supporting the above visualizations.

Open questions

The key open question here is how to efficiently calculate the necessary operation-level statistics for enabling this use-case.

Contributor guide

Tech stack
goreacttypescript
Domain
full stackobservabilityperformance
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
over 1 week
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
fresh
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
GoReactdistributed tracing conceptsstatistics/percentiles
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
30
Research direction
The issue proposes augmenting the Jaeger Trace Timeline with span level aggregate statistics like percentile rankings and latency distribution histograms. The main challenge is efficiently computing operation level statistics from collected traces. To contribute, first review the existing Jaeger frontend code (likely in the jaeger ui repository) to understand the Trace Timeline component, and the backend storage/query layer for span data. The mockups provide clear UI direction. The open question about efficient calculation of percentile rankings needs investigation; exploring sampling or approximation methods is a good starting point. No linked PRs yet.