HIOS: A host interface I/O scheduler for Solid State Disks

Boosting performance of SSD with chip-level RAID by deferring garbage collection

“…To address this issue, Jung et al [4] proposes a host interface I/O scheduler that is both GC-aware and QoS-aware. The scheduler re-distributes GC overheads across non-critical requests to reduce worst-case latency in SSDs.…”

“…During GC, besides erase operation, many unexpected reads and writes increase rapidly due to valid page movement. This causes sharp degradation of SSDs' performance [4]. Fig.…”

“…As analyzed for W 0 , W 2 also chooses RCW to compose new parity groups and it has 1 read and 3 write operations on chip 0 to chip 3. Note that, W 2 on chip 0 would invoke GC, and the response time to W 2 would get a huge increase [1,4] due to the erase operation and valid page movement. Although there are idle chips (chip 4 and chip 5) for W 2 , RAID-5 waits for chip 0 to serve it.…”

“…This results in long waiting time for requests that access the same chips. There exist some recent works [3,4] on optimizing the performance of user requests which are affected by GC, and they achieve good results. There is a new trend that SSDs are deployed with chip-level RAID-5 [3,5,6,7,8], to ensure fault tolerance, as MLC/TLC flash memory technology increases SSDs' capacity but sacrifices their reliability [6,9].…”

“…Traditional approaches of GC have focused on reducing GC execution time [10], exploring various kinds of victim block selection algorithm [11], or hiding GC latency by performing it at non-critical request [4] to realize optimization of GC workflow. However, almost all works require lots of modifications of system.…”