Public‎ > ‎Software Traces‎ > ‎

CERN EOS File System Traces

Location: CERN
Tool: 
Year: 2016-2017
Summary:

This is a set of file access traces, they span 2.49 billion file accesses spanning over 11 months on the CERN EOS system. A description of field names can be found in the CERN documentation

Reading:

These traces are parquet files which can be read using the Apache Parquet system. Easiest way is to use the parquet parsing from Python. Scripts are available in the traces directory to convert them to CSVs using apache arrow.

Documentation:
Dev Purandare: Trace Analysis of Large Scale Storage Systems, UCSC Master's project report 2019.

Data:
On the CRSS servers, they are available at /data/main-share/cern_data NFS directory
Comments