Latency management in storage systems

Research output: Chapter in Book/Report/Conference proceedingConference contribution

7 Citations (Scopus)

Abstract

Storage Latency Estimation Descriptors, or SLEDs, are an API that allow applications to understand and take advantage of the dynamic state of a storage system. By accessing data in the file system cache or high-speed storage first, total I/O workloads can be reduced and performance improved. SLEDs report estimated data latency, allowing users, system utilities, and scripts to make file access decisions based on those retrieval time estimates. SLEDs thus can be used to improve individual application performance, reduce system workloads, and improve the user experience with more predictable behavior. We have modified the Linux 2.2 kernel to support SLEDs, and several Unix utilities and astronomical applications have been modified to use them. As a result, execution times of the Unix utilities when data file sizes exceed the size of the file system buffer cache have been reduced from 50% up to more than an order of magnitude. The astronomical applications incurred 30-50% fewer page faults and reductions in execution time of 10-35%. Performance of applications which use SLEDs also degrade more gracefully as data file size grows.

Original languageEnglish
Title of host publicationProceedings of the 4th Conference on Symposium on Operating System Design and Implementation, OSDI 2000
PublisherAssociation for Computing Machinery, Inc
Publication statusPublished - 2000 Oct 22
Externally publishedYes
Event4th Conference on Symposium on Operating System Design and Implementation, OSDI 2000 - San Diego, United States
Duration: 2000 Oct 222000 Oct 25

Other

Other4th Conference on Symposium on Operating System Design and Implementation, OSDI 2000
CountryUnited States
CitySan Diego
Period00/10/2200/10/25

Fingerprint

Application programming interfaces (API)
Linux

ASJC Scopus subject areas

  • Computer Science(all)

Cite this

Van Meter, R. D., & Gao, M. (2000). Latency management in storage systems. In Proceedings of the 4th Conference on Symposium on Operating System Design and Implementation, OSDI 2000 Association for Computing Machinery, Inc.

Latency management in storage systems. / Van Meter, Rodney D; Gao, Minxi.

Proceedings of the 4th Conference on Symposium on Operating System Design and Implementation, OSDI 2000. Association for Computing Machinery, Inc, 2000.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Van Meter, RD & Gao, M 2000, Latency management in storage systems. in Proceedings of the 4th Conference on Symposium on Operating System Design and Implementation, OSDI 2000. Association for Computing Machinery, Inc, 4th Conference on Symposium on Operating System Design and Implementation, OSDI 2000, San Diego, United States, 00/10/22.
Van Meter RD, Gao M. Latency management in storage systems. In Proceedings of the 4th Conference on Symposium on Operating System Design and Implementation, OSDI 2000. Association for Computing Machinery, Inc. 2000
Van Meter, Rodney D ; Gao, Minxi. / Latency management in storage systems. Proceedings of the 4th Conference on Symposium on Operating System Design and Implementation, OSDI 2000. Association for Computing Machinery, Inc, 2000.
@inproceedings{c4742403cf64471d943b5456a98a39d0,
title = "Latency management in storage systems",
abstract = "Storage Latency Estimation Descriptors, or SLEDs, are an API that allow applications to understand and take advantage of the dynamic state of a storage system. By accessing data in the file system cache or high-speed storage first, total I/O workloads can be reduced and performance improved. SLEDs report estimated data latency, allowing users, system utilities, and scripts to make file access decisions based on those retrieval time estimates. SLEDs thus can be used to improve individual application performance, reduce system workloads, and improve the user experience with more predictable behavior. We have modified the Linux 2.2 kernel to support SLEDs, and several Unix utilities and astronomical applications have been modified to use them. As a result, execution times of the Unix utilities when data file sizes exceed the size of the file system buffer cache have been reduced from 50{\%} up to more than an order of magnitude. The astronomical applications incurred 30-50{\%} fewer page faults and reductions in execution time of 10-35{\%}. Performance of applications which use SLEDs also degrade more gracefully as data file size grows.",
author = "{Van Meter}, {Rodney D} and Minxi Gao",
year = "2000",
month = "10",
day = "22",
language = "English",
booktitle = "Proceedings of the 4th Conference on Symposium on Operating System Design and Implementation, OSDI 2000",
publisher = "Association for Computing Machinery, Inc",

}

TY - GEN

T1 - Latency management in storage systems

AU - Van Meter, Rodney D

AU - Gao, Minxi

PY - 2000/10/22

Y1 - 2000/10/22

N2 - Storage Latency Estimation Descriptors, or SLEDs, are an API that allow applications to understand and take advantage of the dynamic state of a storage system. By accessing data in the file system cache or high-speed storage first, total I/O workloads can be reduced and performance improved. SLEDs report estimated data latency, allowing users, system utilities, and scripts to make file access decisions based on those retrieval time estimates. SLEDs thus can be used to improve individual application performance, reduce system workloads, and improve the user experience with more predictable behavior. We have modified the Linux 2.2 kernel to support SLEDs, and several Unix utilities and astronomical applications have been modified to use them. As a result, execution times of the Unix utilities when data file sizes exceed the size of the file system buffer cache have been reduced from 50% up to more than an order of magnitude. The astronomical applications incurred 30-50% fewer page faults and reductions in execution time of 10-35%. Performance of applications which use SLEDs also degrade more gracefully as data file size grows.

AB - Storage Latency Estimation Descriptors, or SLEDs, are an API that allow applications to understand and take advantage of the dynamic state of a storage system. By accessing data in the file system cache or high-speed storage first, total I/O workloads can be reduced and performance improved. SLEDs report estimated data latency, allowing users, system utilities, and scripts to make file access decisions based on those retrieval time estimates. SLEDs thus can be used to improve individual application performance, reduce system workloads, and improve the user experience with more predictable behavior. We have modified the Linux 2.2 kernel to support SLEDs, and several Unix utilities and astronomical applications have been modified to use them. As a result, execution times of the Unix utilities when data file sizes exceed the size of the file system buffer cache have been reduced from 50% up to more than an order of magnitude. The astronomical applications incurred 30-50% fewer page faults and reductions in execution time of 10-35%. Performance of applications which use SLEDs also degrade more gracefully as data file size grows.

UR - http://www.scopus.com/inward/record.url?scp=84956862916&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84956862916&partnerID=8YFLogxK

M3 - Conference contribution

BT - Proceedings of the 4th Conference on Symposium on Operating System Design and Implementation, OSDI 2000

PB - Association for Computing Machinery, Inc

ER -