In order to efficiently differentiate the amount of written data
to each SSD, full stripe writes should be converted to partial
stripe writes. However, a naive solution such as using a large
chunk size significantly degrades the lifetime of SSDs because
of a large increase in parity updates (as explained in Section
II). In this paper, we propose a new lifetime management
technique, DA-RAID, for RAID systems which overcomes the
technical limitation of Diff-RAID without degrading the SSD
lifetime. In DA-RAID, deduplication is employed as a main instrument
of converting full stripe writes to partial stripe writes.
By removing duplicated pages from a full stripe write using
a deduplication technique, we can convert many full stripe
writes into partial stripe writes. Since the converted partial
stripe writes enable more flexibility in deciding a destination
SSD for each page of partial stripe writes we can better meet
the age difference requirement for among SSDs. In order to
sustain the required age differences among SSDs, we also
propose a simple but effective SSD re-allocation technique that
adaptively changes destination SSDs for partial stripe writes
by accounting for each SSD’s aging pattern. Our experimental
results show that DA-RAID can effectively sustain the age
difference among SSDs even for data center workloads (for
which Diff-RAID fails) without additional parity updates. DARAID
achieves similar age differences among SSDs as Diff-
RAID with a large chunk size while reducing the amount of
parity updates by 32% over Diff-RAID.
The rest of the paper is organized as follows. Section II
describes the limitations of existing RAID and Diff-RAID
techniques. Section III presents the proposed DA-RAID in
detail and shows how it solves problems with the existing
RAID solutions. In Section IV, experimental results of DARAID
are presented. Finally, Section V concludes with a
summary and future work.