Corrections to the FAST'21 paper "An In-Depth Study of Correlated Failures in
Production SSD-Based Data Centers"

*************************************************************************
January 17, 2022

We fixed a bug in our analysis source code (i.e., the file path "analysis/measure.py").

The bug affected the relative percentages of intra-node and intra-rack failures
with a threshold of one month in Figures 6, 8, 10, 12, and 17. Specifically, we
wrongly computed the relative percentages with a threshold of "30 hours" as those
with a threshold of "one month". Thus, the yellow bars in these figures should be
larger than the current results.

However, this bug does not affect any text and findings in the paper. Hence,
we still keep the USENIX version as is.