The DOI matches its Crossref record; the references it cites are not checked yet.
MRCA: Metric-level Root Cause Analysis for Microservices via Multi-Modal Data
Abstract
Due to the complexity and dynamic nature of large-scale microservice systems, manual troubleshooting is time-consuming and impractical. Therefore, automated Root Cause Analysis (RCA) is essential. However, existing RCA approaches face significant challenges. (1) Multi-modal data (e.g. traces, logs, and metrics) record the status of microservice systems, but most existing RCA approaches rely on single-source data, failing to understand the system fully. (2) Existing RCA approaches ignore the services' anomaly state and their anomaly intensity. (3) The service-level RCAs lack detailed information for quick issue resolution. To tackle these challenges, we propose MRCA, a metric-level RCA approach using multi-modal data. Our key insight is that using multi-modal data allows for a comprehensive understanding of the system, enabling the localization of root causes across more anomaly scenarios. MRCA first utilizes traces and logs to obtain the ranking list of abnormal services based on reconstruction probability. It further builds causal graphs from services with high anomaly probability to discover the order in which abnormal metrics of different services occur. By incorporating a reward mechanism, MRCA terminates the excessive expansion of the causal graph and significantly reduces the time taken for causal analysis. Finally, MRCA can prune the ranking list based on the causal graph and identify metric-level root causes. Experiments on two widely-used microservice benchmarks demo…
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
An agent may identify the right cause and still fail operationally if it cannot show evidence, pass checks, prepare rollback, or leave an audit trail.
“An agent may identify the right cause and still fail operationally if it cannot show evidence, pass checks, prepare rollback, or leave an audit trail [85], [43], [74].”
From Large language models for agentic NetOps and AIOps: Architectures, evaluation, and safety (2027) · cites this paper as [74]
At low rungs, the main damage is wasted time, unsupported advice, and misleading explanations; at high rungs, the main damage is unsafe change, delayed rollback, and hard-to-audit action.
“At low rungs, the main damage is wasted time, unsupported advice, and misleading explanations.At high rungs, the main damage is unsafe change, delayed rollback, and hard-to-audit action [99], [74].”
From Large language models for agentic NetOps and AIOps: Architectures, evaluation, and safety (2027) · cites this paper as [74]
Gating and verification are binding constraints once T_execute is non-empty.
“The qualitative takeaway is that gating and verification are not optional extras at higher rungs.They are binding constraints once T execute k is non-empty [31], [74].”
From Large language models for agentic NetOps and AIOps: Architectures, evaluation, and safety (2027) · cites this paper as [74]
If an LLM summary cannot be traced back to log lines and timestamps, it becomes a familiar operational nuisance: neat enough to circulate, and unsafe when someone has to act on it.
“If an LLM summary cannot be traced back to log lines and timestamps, it becomes a familiar operational nuisance: neat enough to circulate, and unsafe when someone has to act on it [80], [74].”
From Large language models for agentic NetOps and AIOps: Architectures, evaluation, and safety (2027) · cites this paper as [74]
Recent RCA systems that use LLM tend to require that explanations are based on actual diagnostic evidence.
“Recent RCA systems that use LLM tend to require that explanations are based on actual diagnostic evidence.”
From Large language models for agentic NetOps and AIOps: Architectures, evaluation, and safety (2027) · cites this paper as [74]
Citations
The citation lists haven’t been fetched yet
References 43 · Cited by 18
Research works printed but not linked: 46
- 10.1109/ICSE.2019.00031 · DOI
- 10.1145/357830.357849 · DOI
- 10.1109/MS.2016.64 · DOI · Paper page
- 10.1109/TPAMI.2018.2798607 · DOI · Paper page
- 10.1109/INFOCOM.2014.6848128 · DOI
- 10.1145/3368089.3417055 · DOI · Paper page
- 10.1109/ICSA.2017.24 · DOI
- 10.1145/3611643.3613864 · DOI
- 10.1145/3445814.3446700 · DOI
- 10.1109/MCSE.2018.2873866 · DOI
- Clive WJ Granger. 1969. Investigating causal relations by econometric models and cross-spectral methods. Econometrica: Journal of the Econometric Society (1969), 424--438.
- 10.1109/SCC.2009.48 · DOI
- 10.1109/TMLCN.2024.3409200 · DOI
- 10.1145/1592568.1592597 · DOI
- Diederik P Kingma and Max Welling. 2013. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114 (2013). · arXiv
- 10.1186/s13677-022-00296-4 · DOI
- 10.1109/IWQOS52092.2021.9521340 · DOI · Paper page
- Timothy P Lillicrap, Jonathan J Hunt, Alexander Pritzel, Nicolas Heess, Tom Erez, Yuval Tassa, David Silver, and Daan Wierstra. 2015. Continuous control with deep reinforcement learning. arXiv preprint arXiv:1509.02971 (2015). · arXiv
- 10.1007/978-3-030-03596-9_1 · DOI
- 10.1109/ICSE-SEIP52600.2021.00043 · DOI · Paper page
- Chang Lou, Peng Huang, and Scott Smith. 2020. Understanding, detecting and localizing partial failures in large system software. In 17th USENIX Symposium on Networked Systems Design and Implementation (NSDI 20). 559--574.
- 10.1145/3472883.3487003 · DOI
- 10.1109/ICWS.2019.00022 · DOI · Paper page
- 10.1109/IWQoS49365.2020.9213058 · DOI
- 10.3390/make1010019 · DOI
- Claus Pahl and Pooyan Jamshidi. 2016. Microservices: A Systematic Mapping Study. CLOSER (1) (2016), 137--146.
- 10.1145/3660805 · DOI
- 10.1145/3308558.3313653 · DOI
- Yuta Shimono, Masataka Hakamada, and Mamoru Mabuchi. 2024. NPEX: Never give up protein exploration with deep reinforcement learning. Journal of Molecular Graphics and Modelling (2024), 108802.
- Shreshth Tuli, Giuliano Casale, and Nicholas R Jennings. 2022. Tranad: Deep transformer networks for anomaly detection in multivariate time series data. arXiv preprint arXiv:2201.07284 (2022). · arXiv
- 10.1109/ColumbianCC.2015.7333476 · DOI
- 10.1007/s10115-024-02122-6 · DOI
- 10.1109/CCGRID.2018.00076 · DOI · Paper page
- Li Wu, Jasmin Bogatinovski, Sasho Nedelkoski, Johan Tordsson, and Odej Kao. 2020. Performance diagnosis in cloud microservices using deep learning. In International Conference on Service-Oriented Computing. 85--96.
- 10.1145/3442381.3449905 · DOI · Paper page
- Guangba Yu Pengfei Chen Yufeng Li Hongyang Chen Xiaoyun Li and Zibin Zheng. 2023. Nezha: Interpretable Fine-Grained Root Causes Analysis for Microservices on Multi-modal Observability Data. In The 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 553--565.
- 10.1145/3510003.3510180 · DOI
- 10.1609/aaai.v33i01.33011409 · DOI
- 10.1016/j.aap.2024.107645 · DOI
- 10.1145/3580305.3599902 · DOI
- 10.1007/s11265-021-01644-4 · DOI
- Lecheng Zheng, Zhengzhang Chen, Jingrui He, and Haifeng Chen. 2024. Multimodal Causal Structure Learning and Root Cause Analysis. arXiv preprint arXiv:2402.02357 (2024). · arXiv
- 10.1109/TSE.2018.2887384 · DOI · Paper page
- Ziyuan Zhou, Guanjun Liu, and Ying Tang. 2024. Multiagent Reinforcement Learning: Methods, Trustworthiness, Applications in Intelligent Vehicles, and Challenges. IEEE Transactions on Intelligent Vehicles (2024).
- Shengyu Zhu, Ignavier Ng, and Zhitang Chen. 2019. Causal discovery with reinforcement learning. arXiv preprint arXiv:1906.04477 (2019). · arXiv
- Zhouruixing Zhu, Cheryl Lee, Xiaoying Tang, and Pinjia He. 2024. HeMiRCA: Fine-Grained Root Cause Analysis for Microservices with Heterogeneous Data Sources. ACM Transactions on Software Engineering and Methodology (2024).
Metadata
- Authors
- Wang, Y., Zhu, Z., Fu, Q., Ma, Y., & He, P.
- Affiliations
- Chinese University of Hong Kong, Shenzhen (CN)
- Huawei Technologies (China) (CN)
- Shenzhen Research Institute of Big Data (CN)
- Venue
- Proceedings of the 39th IEEE/ACM International Conference on Automated Software Engineering
- Publisher
- ACM
- Published
- 2024-10-27
- Pages
- 1057-1068
- OpenAlex
- W4403536919
- Cited by
- 18
- References
- 43
- Citations by year
- 2026: 14 · 2025: 4
- FWCI
- 7.88
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- IEEE/ACM International Conference on Automated Software Engineering (ASE) (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Networks and Communications › Software System Performance and Reliability
- Physical Sciences › Computer Science › Information Systems › Cloud Computing and Resource Management
- Physical Sciences › Computer Science › Information Systems › Software Engineering Research
- Keywords
- root cause analysis, microservices, multimodal data, anomaly localization, causal graphs, system logs, automated troubleshooting
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
