Are GNNs Effective for Multimodal Fault Diagnosis in Microservice Systems?
Fei Gao 1 , Ruyue Xin 2 , Yaqiang Zhang 1 · 2025
Read Are GNNs Effective for Multimodal Fault Diagnosis in Microservice Systems? by Fei Gao 1 , Ruyue Xin 2 , Yaqiang Zhang 1 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-02766.txt.
Description
Graph neural networks are the default tool for finding faults in microservice systems, on the theory that they model how services depend on each other. Fei Gao and co-authors built a deliberately graph-free rival to test that theory. Their baseline, DiagMLP, fuses logs, metrics, and traces like its rivals but ignores topology entirely.
Across five datasets, the simple model matched state-of-the-art graph methods at detecting, locating, and classifying faults. The authors argue the heavy lifting was never the graph. Preprocessing already smuggles dependency information into the features, and the graph modules add little beyond multimodal fusion itself.
Catalog record
| CREATOR | Fei Gao 1 , Ruyue Xin 2 , Yaqiang Zhang 1 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 22 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-02766