Optimal Read-Once Parallel Disk Scheduling

Kallahalla, Mahesh; Varman, Peter

doi:10.1007/s00453-004-1129-7

Cited by 7 publications

(3 citation statements)

References 32 publications

(37 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Peter J. Varman etc. proposed other parallel prefetching algorithms [23,24] for parallel disks systems, where they are based on the large scale blocks lookahead or off-line knowledge and try to fetch a block from each disk on every I/O period time. However, the complete future reference string used in these approaches is non-trivial to be achieved in practice.…”

Section: B Block-level Prefetchingmentioning

confidence: 99%

LSP: A Locality-Aware Strip Prefetching Scheme for Striped Disk Array Systems with Concurrent Accesses

Shi¹,

Feng²

2012

JCP

View full text Add to dashboard Cite

In striped disk array systems, the independency of disks for prefetching is more important than parallelism under high concurrency of accesses, based on which strip prefetching with low read cost has more ability to improve the performance of RAID. However, it indiscriminately fetching all the involved strips limits its applicability. To solve this problem, we propose a Locality-aware Strip Prefetching Scheme (LSP), where it keeps track of the users’ accesses and identifies the hot data areas. Only those strips located in the hot data areas will be prefetched and each prefetching request fetches one strip. LSP has several advantages. First, LSP adapts to the evolving workloads in an online and self-tuning fashion and satisfies the striped disk array systems due to the low read cost in each prefetching request. Second, LSP fully exploits the spatial locality in users’ accesses. Here, the spatial locality is more general, which includes multiple simple patterns, such as loop references, sequentiality, reverse references, and other locality-awarded patterns. Third, LSP discriminates the hot data areas from the cold data areas when prefetching, which significantly alleviates the waste of disk bandwidth, optimizes the cache utilization and improves the prefetching accuracy. We have implemented the prototype of LSP algorithm in Linux kernel 2.6.18. The experimental results show that LSP outperforms SP and Sequential prefetching (SEQP) by up to 22.4% and 24.1% in terms of the average response time, and by up to 1.5 times and 2.3 times in terms of throughput, respectively

show abstract

Section: B Block-level Prefetchingmentioning

confidence: 99%

LSP: A Locality-Aware Strip Prefetching Scheme for Striped Disk Array Systems with Concurrent Accesses

Shi¹,

Feng²

2012

JCP

View full text Add to dashboard Cite

show abstract

“…Belady [7] solved the caching problem for a single disk. Kallahalla and Varman [15,17] developed an optimal parallel disk prefetching algorithm for read-once sequences. Besides a simpler algorithm and analysis, our contribution is a linear time algorithm (which is, however, also implicit in [16], which was published simultaneously with the first announcement of our result [14].)…”

mentioning

confidence: 99%

“…Albers [2] showed for a single disk that a lookahead for the next Ω (M/B) different blocks is needed to get good competitiveness. For parallel disks, Kallahalla and Varman [15,17] showed that another factor of Ω (D) lookahead is needed even for the read-once problem. Applying the optimal offline algorithm to the lookahead matches this lower bound for read-once sequences.…”

mentioning

confidence: 99%

Duality Between Prefetching and Queued Writing with Parallel Disks

Hutchinson¹,

Sanders²,

Vitter³

2005

SIAM J. Comput.

View full text Add to dashboard Cite

Parallel disks promise to be a cost effective means for achieving high bandwidth in applications involving massive data sets, but algorithms for parallel disks can be difficult to devise. To combat this problem, we define a useful and natural duality between writing to parallel disks and the seemingly more difficult problem of prefetching. We first explore this duality for applications involving read-once accesses using parallel disks. We get a simple linear time algorithm for computing optimal prefetch schedules and analyze the efficiency of the resulting schedules for randomly placed data and for arbitrary interleaved accesses to striped sequences. Duality also provides an optimal schedule for prefetching plus caching, where blocks can be accessed multiple times. Another application of this duality gives us the first parallel disk sorting algorithms that are provably optimal up to lower-order terms. One of these algorithms is a simple and practical variant of multiway mergesort, addressing a question that had been open for some time.1. Introduction. External memory (EM) algorithms are those for which the problem data set is too large to fit into the high-speed random access memory (RAM) of a computer and therefore must reside on external devices such as disk drives [23]. In order to cope with the high latency of accessing data on disks, efficient EM algorithms exploit locality in their design. In the I/O model, EM algorithms access a large block of B contiguous data elements in one I/O step and perform the necessary algorithmic operations on the elements in the block while in the high-speed memory. The speedup can be significant. However, even with blocked access, a single disk provides much less bandwidth than the internal memory. This problem can be mitigated by using multiple disks in parallel. For each input/output operation, one block is transferred between a fast memory of size M and each of the D disks. The algorithm therefore transfers D blocks at the cost of a single-disk access delay.A simple approach to algorithm design for parallel disks is to employ large logical blocks, or superblocks, of size B · D in the algorithm. This reduces the problem to designing an EM algorithm for one disk with logical block size BD. A superblock is split into D physical blocks-one on each disk. All D physical blocks are accessed *

show abstract

External Sorting and Permuting

Vitter¹

2008

Encyclopedia of Algorithms

View full text Add to dashboard Cite

Optimal Read-Once Parallel Disk Scheduling

Cited by 7 publications

References 32 publications

LSP: A Locality-Aware Strip Prefetching Scheme for Striped Disk Array Systems with Concurrent Accesses

LSP: A Locality-Aware Strip Prefetching Scheme for Striped Disk Array Systems with Concurrent Accesses

Duality Between Prefetching and Queued Writing with Parallel Disks

External Sorting and Permuting

Contact Info

Product

Resources

About