Xinda | Yuxuan's Homepage

Distributed systems are routinely tested for crash faults, but slow faults — nodes that respond correctly but unusually slowly — are far harder to handle. Existing timeout-based defenses are brittle: too aggressive and they cause false positives; too lenient and they let slow nodes stall entire operations.

Xinda is a study and toolkit for characterizing and improving slow-fault tolerance. We conducted a systematic study of how 10 widely-used distributed systems handle slow faults, revealing that one-size-fits-all timeout strategies consistently fail. Xinda then provides adaptive mechanisms that tune fault-tolerance behavior based on runtime signals.

Key contributions:

First large-scale empirical study of slow-fault handling across production distributed systems
Characterization of failure modes specific to slow faults vs. crash faults
Adaptive slow-fault tolerance mechanisms that outperform static timeout policies

(Lu et al., 2025)

Published at NSDI ‘25. Code available on GitHub.

References

2025

One-Size-Fits-None: Understanding and Enhancing Slow-Fault Tolerance in Modern Distributed Systems

Ruiming Lu, Yunchi Lu, Yuxuan Jiang, and 2 more authors

In Proceedings of the 22nd USENIX Symposium on Networked Systems Design and Implementation, Philadelphia, PA, USA, Apr 2025

Bib PDF Code Poster Slides

@inproceedings{SlowFaultStudy2025NSDI,
  author = {Lu, Ruiming and Lu, Yunchi and Jiang, Yuxuan and Xue, Guangtao and Huang, Peng},
  title = {One-Size-Fits-None: Understanding and Enhancing Slow-Fault Tolerance in Modern Distributed Systems},
  booktitle = {Proceedings of the 22nd USENIX Symposium on Networked Systems Design and Implementation},
  series = {NSDI '25},
  month = apr,
  year = {2025},
  location = {Philadelphia, PA, USA},
  bibtex_show = true,
}