The Tale of Errors in Microservices (Artifact part 1)
<p>This dataset provides comprehensive microservice traces (around 1.4 million) collected from Uber microservice architecture, as described in our paper <strong>The Tale of Errors in Microservices</strong>, presented at <strong>SIGMETRICS 2025</strong>. This dataset enables researchers to study microservice behaviors, optimize performance, and investigate latency reduction techniques.</p> <p>The data has been sanitized to protect proprietary information while retaining critical performance characteristics for academic research.</p> <h3>Artifact Structure and Decompression Instructions:</h3> <p>Due to Zenodo's file size constraints and upload issues, the large <code>trace1-sanitized.tar.zst</code> and <code>trace2-sanitized.tar.zst</code> files have been split into multiple pieces. The artifact is available in two parts (<a href="#" rel="noopener">10.5281/zenodo.13947828</a> and <a href="https://doi.org/10.5281/zenodo.13952897">10.5281/zenodo.13952897</a>). To access the sanitized microservice traces, download all the split parts from both artifacts. After downloading, reassemble the files using the following commands and then decompress the <code>.zst</code> files individually. Each <code>.zst</code> file will require 300-500GB of disk space to decompress.</p> <h4>Reassembling the split files:</h4> <p>For <code>trace1-sanitized.tar.zst</code> and <code>trace2-sanitized.tar.zst</code>, use the following commands to reassemble them:</p> <div> <div><code>cat trace1_* > trace1-sanitized.tar.zst</code></div> <div><code>cat trace2_* > trace2-sanitized.tar.zst</code></div> </div> <p>Once reassembled, you can decompress the files:</p> <div> <div><code>zstd -d trace1-sanitized.tar.zst </code></div> <div><code>zstd -d trace2-sanitized.tar.zst </code></div> </div> <h4>Contents of the Traces:</h4> <ul> <li><code>trace1-sanitized.tar.zst</code> and <code>trace2-sanitized.tar.zst</code> (in <a href="#" rel="noopener">10.5281/zenodo.13952897</a>) contain around 1.4 million traces that correspond to the data described in Sections 3 and 4 of the original paper. <strong>Note:</strong> The traces in this dataset were collected on different days than those used in the paper, so analysis results may vary slightly from what is reported in the publication.</li> <li><code>driver-sanitized.tar.zst</code> contains the sanitized version of the original trace and corresponds to the <strong>App-Launch Use Case</strong> discussed in Section 6.3 and Figure 17 of the original paper.</li> </ul> <p><strong>Note</strong></p> <ul> <li>Due to privacy and security concerns, most unrelated fields and tags are removed. However, <strong>error-related</strong> tags are retained.</li> <li>All trace is sanitized consistently. The same service or endpoint will have identical mapping across three directories. (i.e., service 1 represents the same service in all traces). However, the mapping is <strong>inconsistent</strong> with <a href="https://zenodo.org/records/13956078">https://zenodo.org/records/13956078</a>, so please do not mix the traces between the two artifacts.</li> <li>To preserve privacy, the start time of each trace has been randomly shifted. As a result, the start and end times in the traces do not reflect the actual collection times, and users should not attempt to infer when the traces were gathered.</li> <li>Within each trace, the relative durations and timestamps of all spans remained consistent, as the shift was applied uniformly across the entire trace.</li> </ul> <p>If you use the traces in your research, please cite our paper</p> <blockquote> <p><strong>The Tale of Errors in Microservices</strong><br><em>I-Ting Angelina Lee (Washington University in St. Louis); Zhizhou Zhang, Abhishek Parwal (Uber Technologies Inc.); Milind Chabbi (Uber Technologies)<br>SIGMETRICS 2025 <a href="https://doi.org/10.1145/3700436">https://doi.org/10.1145/3700436</a></em></p> </blockquote> <p>If you have more questions, you can reach out to <a href="https://lollllcat.github.io/">Chris(Zhizhou) Zhang</a>.</p>
ShareScore
24/100
Overall dataset sharing score
Score breakdown
These five areas show where the dataset supports — or may limit — practical reuse.
- Stewardship
- 4
- Harmonization
- 4
- Access
- 16
- Reuse readiness
- 0
- Engagement
- 0
ScienceDex for faculty
Your academic CV and NIH biosketch
Make your CV in your school's format and your NIH biosketch from it, then download both as Word files.