P55: Incorporating Proactive Data Rescue into ZFS Disk
Recovery for Enhanced Storage Reliability
SessionPoster Reception
Authors
Event Type
ACM Student Research Competition
Poster
Reception
TimeTuesday, November 14th5:15pm -
7pm
LocationFour Seasons Ballroom
DescriptionAs tremendous amount of data are generated every day,
storage systems store exabytes of data on hundreds of
thousands of disk drives. At such a scale, disk failures
become the norm. Data recovery takes longer time due to
increased disk capacity. ZFS is a widely used
filesystem, providing data recovery from corruption.
Many factors may affect ZFS's recovery performance in a
production environment. Additionally, disk failure
prediction techniques enables ZFS to proactively rescue
data prior to disk failures. In this poster, we
extensively evaluate the recovery performance with a
variety of ZFS configurations. We also compare the
performance of different data rescue strategies,
including post-failure disk recovery, proactive disk
cloning, and proactive data recovery. Our proposed
analytic model uses the collected zpool utilization data
and system configuration to derive the optimal data
rescue strategy that best suits the storage array in the
current state.
