Profit data caching and hybrid disk-aware Completely Fair Queuing scheduling algorithms for hybrid disks

Abstract

SUMMARYRecently, a hybrid disk drive that integrates a small amount of flash memory within a mechanical drive has received significant attention. The hybrid drive extends the storage hierarchy by using flash memory to cache data from the mechanical disk. Unfortunately, current caching architectures fail to fully exploit the potential of the hybrid drive. Furthermore, current disk input/output (I/O) schedulers are optimized for rotational mechanical disk drives and thus must be re-targeted for the hybrid disk drive. In this paper, we propose a new data caching scheme, called Profit Caching, for hybrid drives. Profit Caching is a self-optimizing caching algorithm. It considers and seamlessly integrates all possible data characteristics that impact the performance of hybrid drives, including read count, write count, sequentiality, randomness, and recency, to determine the caching policy. Moreover, we propose a hybrid disk-aware Completely Fair Queuing (HA-CFQ) scheduler to avoid unnecessary I/O anticipations of the CFQ scheduler. We have implemented Profit Caching and HA-CFQ scheduler in the Linux kernel. Coupled with a trace-driven simulator, we have also conducted detailed experiments under a variety of workloads. Experimental results show that Profit Caching provides significantly improved performance compared with the previous schemes. In particular, the throughput of Profit Caching outperforms previous Random Access First and FlashCache caching schemes by factors of up to 1.8 and 7.6, respectively. In addition, the HA-CFQ scheduler reduces the total execution time of the CFQ scheduler by up to 1.74%. Finally, the experimental results show that the runtime overhead of Profit Caching is extremely insignificant and can be ignored.

Host managed contention avoidance storage solutions for Big Data

“…While SLASSD [7] uses an opportunistic goal oriented block I/O scheduling algorithm, Kim et al [2] proposes host level SSD I/O schedulers, which are extensions of state-of-the-art I/O scheduling scheme CFQ. ParDispatcher [49] tries to utilize the parallelism in SSDs, by dividing the entire SSD into sub-regions, each having [6] , SLASSD [7], Axboe [44], Hystor [34], PDC [43], ParDispatcher [49], BFQ [50], FlexDrive [12], AD [28], Borg [8] ADLAM [33], SUORA [4], hatS [21], PDC [43], HRO [45], RPAC [46], RAF [47], PASS [48], Triple-H [11], ExaPlan [20], HybridStore [51], Hystor [34], Scarlett [52], DUX [1] a different queue for dispatching requests. ParDipatcher might be good in applications which have more random I/Os otherwise, leading to increasing wait queues for popular sub-regions and bias in performance.…”

“…HRO [45] migrates or allocates files to SSD based on hotness (access frequency), randomness and profit-value based on read/write-intensiveness and recency of file access. PDC [43] keeps blocks in SSD with highest hit frequency. Migration is based on utility value associated with every block in SSD in last time slot based on read/write counts, known as profit caching.…”

NVMRA: utilizing NVM to improve the random write operations for NAND-flash-based mobile devices

“…">NAND flash memory and NVMIn mobile devices, the performance of NAND flash memory is the major performance bottleneck in I/O storage system. NAND flash memory has several constraints, such as erase-before-rewrite, limited lifetime, and long-time overhead for garbage collection [20][21][22][23][24][25][26][27][28]. All these constrains seriously Figure 2.…”

Hybrid system analysis and design for scale-out storage environments

“…To enhance the overall performance of a storage system, SSD is usually utilized as a cache between the DRAM and HDDs to keep the hot read data or the write data [38,50,55,[62][63][64][65]. Some storage systems with SSD caching methods are shown in Table 2.2.…”

“…The combination of HDD scheduling and SSD scheduling algorithms can be one of the solutions for hybrid storage system. In [144] are two approaches to deal with the data recording to the device managed SMR drives: read-modify-write approach and log-structured recording approach.…”