Geting more performance with polymorphism from emerging memory technologies

Iyswarya Narayanan, Aishwarya Ganesan, Anirudh Badam, Sriram Govindan, Bikash Sharma, Anand Sivasubramaniam

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

Abstract

Storage-intensive systems in data centers rely heavily on DRAM and SSDs for the performance of reads and persistent writes, respectively. These applications pose a diverse set of requirements, and are limited by ixed capacity, ixed access latency, and ixed function of these resources as either memory or storage. In contrast, emerging memory technologies like 3D-Xpoint, battery-backed DRAM, and ASIC-based fast memory-compression ofer capabilities across several dimensions. However, existing proposals to use such technologies can only improve either read or write performance but not both without requiring extensive changes to the application, and the operating system. We present PolyEMT, a system that employs an emerging memory technology based cache to the SSD, and transparently morphs the capabilities of this cache across several dimensions ś persistence, capacity, latency ś to jointly improve both read and write performance. We demonstrate the beneits of PolyEMT using several large-scale storage-intensive workloads from our datacenters.

Original languageEnglish (US)
Title of host publicationSYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference
PublisherAssociation for Computing Machinery, Inc
Pages8-20
Number of pages13
ISBN (Electronic)9781450367493
DOIs
StatePublished - May 22 2019
Event12th ACM International Systems and Storage Conference, SYSTOR 2019 - Haifa, Israel
Duration: Jun 3 2019Jun 5 2019

Publication series

NameSYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference

Conference

Conference12th ACM International Systems and Storage Conference, SYSTOR 2019
CountryIsrael
CityHaifa
Period6/3/196/5/19

Fingerprint

Polymorphism
Data storage equipment
Dynamic random access storage
Application specific integrated circuits

All Science Journal Classification (ASJC) codes

  • Hardware and Architecture
  • Electrical and Electronic Engineering
  • Computer Science Applications
  • Software

Cite this

Narayanan, I., Ganesan, A., Badam, A., Govindan, S., Sharma, B., & Sivasubramaniam, A. (2019). Geting more performance with polymorphism from emerging memory technologies. In SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference (pp. 8-20). (SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference). Association for Computing Machinery, Inc. https://doi.org/10.1145/3319647.3325826
Narayanan, Iyswarya ; Ganesan, Aishwarya ; Badam, Anirudh ; Govindan, Sriram ; Sharma, Bikash ; Sivasubramaniam, Anand. / Geting more performance with polymorphism from emerging memory technologies. SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference. Association for Computing Machinery, Inc, 2019. pp. 8-20 (SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference).
@inproceedings{a69393e4a3de4f7ca3c2ec3f6ae4332e,
title = "Geting more performance with polymorphism from emerging memory technologies",
abstract = "Storage-intensive systems in data centers rely heavily on DRAM and SSDs for the performance of reads and persistent writes, respectively. These applications pose a diverse set of requirements, and are limited by ixed capacity, ixed access latency, and ixed function of these resources as either memory or storage. In contrast, emerging memory technologies like 3D-Xpoint, battery-backed DRAM, and ASIC-based fast memory-compression ofer capabilities across several dimensions. However, existing proposals to use such technologies can only improve either read or write performance but not both without requiring extensive changes to the application, and the operating system. We present PolyEMT, a system that employs an emerging memory technology based cache to the SSD, and transparently morphs the capabilities of this cache across several dimensions ś persistence, capacity, latency ś to jointly improve both read and write performance. We demonstrate the beneits of PolyEMT using several large-scale storage-intensive workloads from our datacenters.",
author = "Iyswarya Narayanan and Aishwarya Ganesan and Anirudh Badam and Sriram Govindan and Bikash Sharma and Anand Sivasubramaniam",
year = "2019",
month = "5",
day = "22",
doi = "10.1145/3319647.3325826",
language = "English (US)",
series = "SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference",
publisher = "Association for Computing Machinery, Inc",
pages = "8--20",
booktitle = "SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference",

}

Narayanan, I, Ganesan, A, Badam, A, Govindan, S, Sharma, B & Sivasubramaniam, A 2019, Geting more performance with polymorphism from emerging memory technologies. in SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference. SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference, Association for Computing Machinery, Inc, pp. 8-20, 12th ACM International Systems and Storage Conference, SYSTOR 2019, Haifa, Israel, 6/3/19. https://doi.org/10.1145/3319647.3325826

Geting more performance with polymorphism from emerging memory technologies. / Narayanan, Iyswarya; Ganesan, Aishwarya; Badam, Anirudh; Govindan, Sriram; Sharma, Bikash; Sivasubramaniam, Anand.

SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference. Association for Computing Machinery, Inc, 2019. p. 8-20 (SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference).

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

TY - GEN

T1 - Geting more performance with polymorphism from emerging memory technologies

AU - Narayanan, Iyswarya

AU - Ganesan, Aishwarya

AU - Badam, Anirudh

AU - Govindan, Sriram

AU - Sharma, Bikash

AU - Sivasubramaniam, Anand

PY - 2019/5/22

Y1 - 2019/5/22

N2 - Storage-intensive systems in data centers rely heavily on DRAM and SSDs for the performance of reads and persistent writes, respectively. These applications pose a diverse set of requirements, and are limited by ixed capacity, ixed access latency, and ixed function of these resources as either memory or storage. In contrast, emerging memory technologies like 3D-Xpoint, battery-backed DRAM, and ASIC-based fast memory-compression ofer capabilities across several dimensions. However, existing proposals to use such technologies can only improve either read or write performance but not both without requiring extensive changes to the application, and the operating system. We present PolyEMT, a system that employs an emerging memory technology based cache to the SSD, and transparently morphs the capabilities of this cache across several dimensions ś persistence, capacity, latency ś to jointly improve both read and write performance. We demonstrate the beneits of PolyEMT using several large-scale storage-intensive workloads from our datacenters.

AB - Storage-intensive systems in data centers rely heavily on DRAM and SSDs for the performance of reads and persistent writes, respectively. These applications pose a diverse set of requirements, and are limited by ixed capacity, ixed access latency, and ixed function of these resources as either memory or storage. In contrast, emerging memory technologies like 3D-Xpoint, battery-backed DRAM, and ASIC-based fast memory-compression ofer capabilities across several dimensions. However, existing proposals to use such technologies can only improve either read or write performance but not both without requiring extensive changes to the application, and the operating system. We present PolyEMT, a system that employs an emerging memory technology based cache to the SSD, and transparently morphs the capabilities of this cache across several dimensions ś persistence, capacity, latency ś to jointly improve both read and write performance. We demonstrate the beneits of PolyEMT using several large-scale storage-intensive workloads from our datacenters.

UR - http://www.scopus.com/inward/record.url?scp=85067114099&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85067114099&partnerID=8YFLogxK

U2 - 10.1145/3319647.3325826

DO - 10.1145/3319647.3325826

M3 - Conference contribution

T3 - SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference

SP - 8

EP - 20

BT - SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference

PB - Association for Computing Machinery, Inc

ER -

Narayanan I, Ganesan A, Badam A, Govindan S, Sharma B, Sivasubramaniam A. Geting more performance with polymorphism from emerging memory technologies. In SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference. Association for Computing Machinery, Inc. 2019. p. 8-20. (SYSTOR 2019 - Proceedings of the 12th ACM International Systems and Storage Conference). https://doi.org/10.1145/3319647.3325826