Classifying data to reduce long term data movement in shingled write disks

Abstract

Abstract-Shingled Magnetic Recording (SMR) is a means of increasing the density of hard drives that brings a new set of challenges. Due to the nature of SMR disks, updating in place is not an option. Holes left by invalidated data can only be filled if the entire band is reclaimed, and a poor band compaction algorithm could result in spending a lot of time moving blocks over the lifetime of the device. We propose using write frequency to separate blocks to reduce data movement and develop a band compaction algorithm that implements this heuristic. We demonstrate how our algorithm results in improved data management, resulting in an up to 47% reduction in required data movements when compared to naive approaches to band management.

SMORE: A Cold Data Object Store for SMR Drives (Extended Version)

“…Categorizing data based on hotness can significantly decrease write amplification on SMR drives [31], [35], [36], and has also been found helpful in the Flash-Friendly File System (F2FS) [37].…”

Modeling and improving performance of write out-of-place storage technologies

“…Since the advent of shingled magnetic recording [18], SMR drives have been one of the main topics of many research studies of the storage community. Prior work on shingled magnetic recording has covered a range of topics, from the potential and possibilities of using shingled drives [19,20], translation layers and data management [21,22,23,24,25,26], SMR-specific file systems [27,28,29] and shingled drive characterization [30].…”

“…Jones et. al [24] propose a similar form of segregation based on write frequency, using the write history of blocks to decrease the amount of data moved due to zone compaction or cleaning.…”

Hybrid system analysis and design for scale-out storage environments

“…The NAND flash is utilized as a cache buffer to keep the write data, which minimizes the number of writing updates to the SMR drive. In [36], the write data is classified into hot data and cold data based on the update frequency, and which are utilized to minimize the data movement in GC process. In [37], a dynamic variable-width striping RAID (DVS-RAID) architecture is proposed for SMR drives, and the proposed method can reduce the updating cost of the parity data.…”