Triple-A

A non-SSD based Autonomic All-Flash Array for high performance storage systems

Myoungsoo Jung, Wonil Choi, John Shalf, Mahmut Taylan Kandemir

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

20 Citations (Scopus)

Abstract

Solid State Disk (SSD) arrays are in a position to (as least partially) replace spinning disk arrays in high performance computing (HPC) systems due to their better performance and lower power consumption. However, these emerging SSD arrays are facing enormous challenges, which are not observed in disk-based arrays. Specifically, we observe that the performance of SSD arrays can significantly degrade due to various array-level resource contentions. In addition, their maintenance costs exponentially increase over time, which renders them diff cult to deploy widely in HPC systems. To address these challenges, we propose Triple-A, a non-SSD based Autonomic All-Flash Array, which is a self-optimizing, from-scratch NAND f ash cluster. Triple-A can detect two different types of resource contentions and autonomically alleviate them by reshaping the physical data-layout on its f ash array network. Our experimental evaluation using both real workloads and a micro-benchmark show that Triple-A can offer a 53% higher sustained throughput and a 80% lower I/O latency than non-autonomic SSD arrays

Original languageEnglish
Title of host publicationASPLOS 2014 - 19th International Conference on Architectural Support for Programming Languages and Operating Systems
Pages441-454
Number of pages14
DOIs
Publication statusPublished - 2014 Mar 14
Event19th International Conference on Architectural Support for Programming Languages and Operating Systems, ASPLOS 2014 - Salt Lake City, UT, United States
Duration: 2014 Mar 12014 Mar 5

Publication series

NameInternational Conference on Architectural Support for Programming Languages and Operating Systems - ASPLOS

Other

Other19th International Conference on Architectural Support for Programming Languages and Operating Systems, ASPLOS 2014
CountryUnited States
CitySalt Lake City, UT
Period14/3/114/3/5

Fingerprint

HPSS
Ashes
Electric power utilization
Throughput
Costs

All Science Journal Classification (ASJC) codes

  • Software
  • Information Systems
  • Hardware and Architecture

Cite this

Jung, M., Choi, W., Shalf, J., & Kandemir, M. T. (2014). Triple-A: A non-SSD based Autonomic All-Flash Array for high performance storage systems. In ASPLOS 2014 - 19th International Conference on Architectural Support for Programming Languages and Operating Systems (pp. 441-454). (International Conference on Architectural Support for Programming Languages and Operating Systems - ASPLOS). https://doi.org/10.1145/2541940.2541953
Jung, Myoungsoo ; Choi, Wonil ; Shalf, John ; Kandemir, Mahmut Taylan. / Triple-A : A non-SSD based Autonomic All-Flash Array for high performance storage systems. ASPLOS 2014 - 19th International Conference on Architectural Support for Programming Languages and Operating Systems. 2014. pp. 441-454 (International Conference on Architectural Support for Programming Languages and Operating Systems - ASPLOS).
@inproceedings{908dfa93f516412098b6a2d492d448ca,
title = "Triple-A: A non-SSD based Autonomic All-Flash Array for high performance storage systems",
abstract = "Solid State Disk (SSD) arrays are in a position to (as least partially) replace spinning disk arrays in high performance computing (HPC) systems due to their better performance and lower power consumption. However, these emerging SSD arrays are facing enormous challenges, which are not observed in disk-based arrays. Specifically, we observe that the performance of SSD arrays can significantly degrade due to various array-level resource contentions. In addition, their maintenance costs exponentially increase over time, which renders them diff cult to deploy widely in HPC systems. To address these challenges, we propose Triple-A, a non-SSD based Autonomic All-Flash Array, which is a self-optimizing, from-scratch NAND f ash cluster. Triple-A can detect two different types of resource contentions and autonomically alleviate them by reshaping the physical data-layout on its f ash array network. Our experimental evaluation using both real workloads and a micro-benchmark show that Triple-A can offer a 53{\%} higher sustained throughput and a 80{\%} lower I/O latency than non-autonomic SSD arrays",
author = "Myoungsoo Jung and Wonil Choi and John Shalf and Kandemir, {Mahmut Taylan}",
year = "2014",
month = "3",
day = "14",
doi = "10.1145/2541940.2541953",
language = "English",
isbn = "9781450323055",
series = "International Conference on Architectural Support for Programming Languages and Operating Systems - ASPLOS",
pages = "441--454",
booktitle = "ASPLOS 2014 - 19th International Conference on Architectural Support for Programming Languages and Operating Systems",

}

Jung, M, Choi, W, Shalf, J & Kandemir, MT 2014, Triple-A: A non-SSD based Autonomic All-Flash Array for high performance storage systems. in ASPLOS 2014 - 19th International Conference on Architectural Support for Programming Languages and Operating Systems. International Conference on Architectural Support for Programming Languages and Operating Systems - ASPLOS, pp. 441-454, 19th International Conference on Architectural Support for Programming Languages and Operating Systems, ASPLOS 2014, Salt Lake City, UT, United States, 14/3/1. https://doi.org/10.1145/2541940.2541953

Triple-A : A non-SSD based Autonomic All-Flash Array for high performance storage systems. / Jung, Myoungsoo; Choi, Wonil; Shalf, John; Kandemir, Mahmut Taylan.

ASPLOS 2014 - 19th International Conference on Architectural Support for Programming Languages and Operating Systems. 2014. p. 441-454 (International Conference on Architectural Support for Programming Languages and Operating Systems - ASPLOS).

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

TY - GEN

T1 - Triple-A

T2 - A non-SSD based Autonomic All-Flash Array for high performance storage systems

AU - Jung, Myoungsoo

AU - Choi, Wonil

AU - Shalf, John

AU - Kandemir, Mahmut Taylan

PY - 2014/3/14

Y1 - 2014/3/14

N2 - Solid State Disk (SSD) arrays are in a position to (as least partially) replace spinning disk arrays in high performance computing (HPC) systems due to their better performance and lower power consumption. However, these emerging SSD arrays are facing enormous challenges, which are not observed in disk-based arrays. Specifically, we observe that the performance of SSD arrays can significantly degrade due to various array-level resource contentions. In addition, their maintenance costs exponentially increase over time, which renders them diff cult to deploy widely in HPC systems. To address these challenges, we propose Triple-A, a non-SSD based Autonomic All-Flash Array, which is a self-optimizing, from-scratch NAND f ash cluster. Triple-A can detect two different types of resource contentions and autonomically alleviate them by reshaping the physical data-layout on its f ash array network. Our experimental evaluation using both real workloads and a micro-benchmark show that Triple-A can offer a 53% higher sustained throughput and a 80% lower I/O latency than non-autonomic SSD arrays

AB - Solid State Disk (SSD) arrays are in a position to (as least partially) replace spinning disk arrays in high performance computing (HPC) systems due to their better performance and lower power consumption. However, these emerging SSD arrays are facing enormous challenges, which are not observed in disk-based arrays. Specifically, we observe that the performance of SSD arrays can significantly degrade due to various array-level resource contentions. In addition, their maintenance costs exponentially increase over time, which renders them diff cult to deploy widely in HPC systems. To address these challenges, we propose Triple-A, a non-SSD based Autonomic All-Flash Array, which is a self-optimizing, from-scratch NAND f ash cluster. Triple-A can detect two different types of resource contentions and autonomically alleviate them by reshaping the physical data-layout on its f ash array network. Our experimental evaluation using both real workloads and a micro-benchmark show that Triple-A can offer a 53% higher sustained throughput and a 80% lower I/O latency than non-autonomic SSD arrays

UR - http://www.scopus.com/inward/record.url?scp=84897778859&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84897778859&partnerID=8YFLogxK

U2 - 10.1145/2541940.2541953

DO - 10.1145/2541940.2541953

M3 - Conference contribution

SN - 9781450323055

T3 - International Conference on Architectural Support for Programming Languages and Operating Systems - ASPLOS

SP - 441

EP - 454

BT - ASPLOS 2014 - 19th International Conference on Architectural Support for Programming Languages and Operating Systems

ER -

Jung M, Choi W, Shalf J, Kandemir MT. Triple-A: A non-SSD based Autonomic All-Flash Array for high performance storage systems. In ASPLOS 2014 - 19th International Conference on Architectural Support for Programming Languages and Operating Systems. 2014. p. 441-454. (International Conference on Architectural Support for Programming Languages and Operating Systems - ASPLOS). https://doi.org/10.1145/2541940.2541953