15-loghub-issre23

Loghub: Large Collection of System Log Datasets

• Venue: ISSRE’23
• Authors: Jieming Zhu, Shilin He, Jinyang Liu et al. (logpai team)
• Paper: https://arxiv.org/abs/2008.06448
• Project: https://github.com/logpai/loghub
• Status: Read

Takeaway

• The original loghub: 19 real-world log datasets (distributed systems, supercomputers, OS, mobile, server apps, standalone)
• 90K+ downloads, used by hundreds of orgs — the de facto standard benchmark for log analysis research
• This is the v1; Loghub-2.0 (ref 03) expanded it to 14 larger datasets with avg 3.6M lines each
• For LAPP: use Loghub-2.0 for benchmarking, but loghub v1 (2K samples per dataset) is handy for quick smoke tests during development