Performance Comparison of Mirrored Disk Scheduling Methods with a Shared Non-Volatile Cache

Abstract

Mirrored disks or RAID1 is a popular disk array paradigm, which in addition to fault-tolerance, doubles the data access bandwidth. This is important in view of rapidly increasing disk capacities and the slow improvement in disk access time. Caching of dirty data blocks in a non-volatile storage (NVS) cache allows the destaging of dirty blocks to be deferrable, so as to improve the response time of read requests by giving them a higher priority than write requests. Destaging of dirty blocks in batches to take advantage of disk geometry entails in lowered disk utilization due to writes and improved performance for reads. Polyzois et al. [12] propose a scheduling policy for mirrored disks equipped with an NVS cache, so that one disk processes read requests, while the other disk is processing a write batch according to the CSCAN policy. We propose an improved scheduling policy as follows: (i) eliminating the forced idleness caused by the batch processing paradigm for write requests, i.e., allowing write requests to be processed individually; (ii) using SATF or even an exhaustive search, to reduce destaging time compared to CSCAN; (iii) introducing a threshold for the number of read requests, which when exceeded defers the destaging of dirty blocks. We compare these two scheduling policies with each other and also against prioritizing the processing of reads versus writes: (i) the head-of-the-line (HOL) priority queueing discipline, (ii) SATF with conditional priorities. It follows from simulation results that the new method outperforms Polyzois' method, which is even outperformed by the HOL priority policy. SATF with conditional priorities slightly outperforms the proposed method from the viewpoint of its throughput and response time, but is susceptible to more variability in response time.

Evaluating Reliability of SSD-Based I/O Caches in Enterprise Storage Systems

“…In this section, we evaluate the impact of using SSDs from different vendors in different orders in RAID-1, used in I/O cache configuration, on the ratio of failures. In a mirrored (i.e., RAID-1) configuration with two disks, the first disk is called "primary" and the second is called "secondary" [15][16][17][18][19][20][21]. To provide data consistency in such subsystem, the controller periodically compares the written data in both disks.…”

“…Such configuration improves both performance and reliability but cannot completely resolve the reliability issues at the RAID level. The mirrored configuration keeps two copies of each data in two different devices known as primary and secondary disks (i.e., primary is the one that the RAID controller chooses to write first) and provides high level of reliability upon disk failures [15][16][17][18][19][20][21]. For each write request coming from the application, two identical write operations are performed in both primary and secondary disks.…”

“…RAID-1 configuration doubles the read performance of the disk subsystem by involving only one of the disks which has a minimum queue size and service time for read operations. Such configuration tolerates disk failures in the subsystem while data failures such as shorn writes (i.e., incomplete writes), flying writes (i.e., misplaced writes), and unserializablity (i.e., out-oforder writes) reported in [12][13][14]22] cannot be tolerated completely [15][16][17][18][19][20][21]. However, a process namely inconsistency check or scrubbing runs in the background and regularly checks the consistency of primary and secondary disks in RAID configuration.…”

Analysis of Fork/Join and Related Queueing Systems

“…In turn, reads associated with F/J requests to materialize data on a failed disk can be processed at a higher priority than ordinary reads to equalize disk access times. A scheme that alternates in utilizing mirrored disks for reading and writing is described in Polyzois et al [1993] and further extended in Thomasian and Liu [2005]. When a disk fails in basic mirroring, the rate of read requests to the surviving disk is doubled.…”

Mirrored disk rouing and scheduling

“…Since the processing of writes can be deferred, the two disks can be used alternatively in processing reads and writes, so that one disk is available at all times for reading [18]. A further study shows that this method can be outperformed by some of the simpler methods proposed in this paper [28].…”