LACS: Load-aware cache sharing with isolation guarantee

Yinghao Yu, Wei Wang, Jun Zhang, Khaled Ben Letaief

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

Abstract

Cluster caching has been increasingly deployed in front of cloud storage to improve I/O performance. In shared, multi-tenant environments such as cloud datacenters, cluster caches are constantly contended by many users. Enforcing performance isolation between users hence becomes imperative to cluster caching. A user's caching performance critically depends on two factors: (1) the amount of cache allocation and (2) the load of servers in which its files are cached. However, existing cache sharing policies only provide guarantees on the amount of cache allocation, while remaining agnostic to the load of cache servers. Consequently, 'mice' users having files co-located with 'elephants' contributing heavy data accesses may experience extremely long latency, hence receiving no isolation. In this paper, we propose a Load-Aware Cache Sharing scheme (LACS) to enforce isolation between users. LACS keeps track of the load contributed by each user and reins back the congestions caused by elephant users by throttling their cache usage and network bandwidth. We have implemented LACS atop Alluxio, a popular cluster caching system. EC2 deployment shows that LACS achieves performance isolation in the presence of elephants, while improving the mean read latency by up to 80.4% (25.3% on average) over the state-of-the-art load balancing technique.

Original languageEnglish
Title of host publicationProceedings - 2019 39th IEEE International Conference on Distributed Computing Systems, ICDCS 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages207-217
Number of pages11
ISBN (Electronic)9781728125190
DOIs
Publication statusPublished - Jul 2019
Event39th IEEE International Conference on Distributed Computing Systems, ICDCS 2019 - Richardson, United States
Duration: 7 Jul 20199 Jul 2019

Publication series

NameProceedings - International Conference on Distributed Computing Systems
Volume2019-July

Conference

Conference39th IEEE International Conference on Distributed Computing Systems, ICDCS 2019
CountryUnited States
CityRichardson
Period7/7/199/7/19

Fingerprint

Servers
Resource allocation
Bandwidth

Keywords

  • Fair cache sharing
  • Isolation guarantte

ASJC Scopus subject areas

  • Software
  • Hardware and Architecture
  • Computer Networks and Communications

Cite this

Yu, Y., Wang, W., Zhang, J., & Letaief, K. B. (2019). LACS: Load-aware cache sharing with isolation guarantee. In Proceedings - 2019 39th IEEE International Conference on Distributed Computing Systems, ICDCS 2019 (pp. 207-217). [8885014] (Proceedings - International Conference on Distributed Computing Systems; Vol. 2019-July). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICDCS.2019.00029

LACS : Load-aware cache sharing with isolation guarantee. / Yu, Yinghao; Wang, Wei; Zhang, Jun; Letaief, Khaled Ben.

Proceedings - 2019 39th IEEE International Conference on Distributed Computing Systems, ICDCS 2019. Institute of Electrical and Electronics Engineers Inc., 2019. p. 207-217 8885014 (Proceedings - International Conference on Distributed Computing Systems; Vol. 2019-July).

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

Yu, Y, Wang, W, Zhang, J & Letaief, KB 2019, LACS: Load-aware cache sharing with isolation guarantee. in Proceedings - 2019 39th IEEE International Conference on Distributed Computing Systems, ICDCS 2019., 8885014, Proceedings - International Conference on Distributed Computing Systems, vol. 2019-July, Institute of Electrical and Electronics Engineers Inc., pp. 207-217, 39th IEEE International Conference on Distributed Computing Systems, ICDCS 2019, Richardson, United States, 7/7/19. https://doi.org/10.1109/ICDCS.2019.00029
Yu Y, Wang W, Zhang J, Letaief KB. LACS: Load-aware cache sharing with isolation guarantee. In Proceedings - 2019 39th IEEE International Conference on Distributed Computing Systems, ICDCS 2019. Institute of Electrical and Electronics Engineers Inc. 2019. p. 207-217. 8885014. (Proceedings - International Conference on Distributed Computing Systems). https://doi.org/10.1109/ICDCS.2019.00029
Yu, Yinghao ; Wang, Wei ; Zhang, Jun ; Letaief, Khaled Ben. / LACS : Load-aware cache sharing with isolation guarantee. Proceedings - 2019 39th IEEE International Conference on Distributed Computing Systems, ICDCS 2019. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 207-217 (Proceedings - International Conference on Distributed Computing Systems).
@inproceedings{84f12ba4b89747a3b1718c5b9ab5d817,
title = "LACS: Load-aware cache sharing with isolation guarantee",
abstract = "Cluster caching has been increasingly deployed in front of cloud storage to improve I/O performance. In shared, multi-tenant environments such as cloud datacenters, cluster caches are constantly contended by many users. Enforcing performance isolation between users hence becomes imperative to cluster caching. A user's caching performance critically depends on two factors: (1) the amount of cache allocation and (2) the load of servers in which its files are cached. However, existing cache sharing policies only provide guarantees on the amount of cache allocation, while remaining agnostic to the load of cache servers. Consequently, 'mice' users having files co-located with 'elephants' contributing heavy data accesses may experience extremely long latency, hence receiving no isolation. In this paper, we propose a Load-Aware Cache Sharing scheme (LACS) to enforce isolation between users. LACS keeps track of the load contributed by each user and reins back the congestions caused by elephant users by throttling their cache usage and network bandwidth. We have implemented LACS atop Alluxio, a popular cluster caching system. EC2 deployment shows that LACS achieves performance isolation in the presence of elephants, while improving the mean read latency by up to 80.4{\%} (25.3{\%} on average) over the state-of-the-art load balancing technique.",
keywords = "Fair cache sharing, Isolation guarantte",
author = "Yinghao Yu and Wei Wang and Jun Zhang and Letaief, {Khaled Ben}",
year = "2019",
month = "7",
doi = "10.1109/ICDCS.2019.00029",
language = "English",
series = "Proceedings - International Conference on Distributed Computing Systems",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "207--217",
booktitle = "Proceedings - 2019 39th IEEE International Conference on Distributed Computing Systems, ICDCS 2019",

}

TY - GEN

T1 - LACS

T2 - Load-aware cache sharing with isolation guarantee

AU - Yu, Yinghao

AU - Wang, Wei

AU - Zhang, Jun

AU - Letaief, Khaled Ben

PY - 2019/7

Y1 - 2019/7

N2 - Cluster caching has been increasingly deployed in front of cloud storage to improve I/O performance. In shared, multi-tenant environments such as cloud datacenters, cluster caches are constantly contended by many users. Enforcing performance isolation between users hence becomes imperative to cluster caching. A user's caching performance critically depends on two factors: (1) the amount of cache allocation and (2) the load of servers in which its files are cached. However, existing cache sharing policies only provide guarantees on the amount of cache allocation, while remaining agnostic to the load of cache servers. Consequently, 'mice' users having files co-located with 'elephants' contributing heavy data accesses may experience extremely long latency, hence receiving no isolation. In this paper, we propose a Load-Aware Cache Sharing scheme (LACS) to enforce isolation between users. LACS keeps track of the load contributed by each user and reins back the congestions caused by elephant users by throttling their cache usage and network bandwidth. We have implemented LACS atop Alluxio, a popular cluster caching system. EC2 deployment shows that LACS achieves performance isolation in the presence of elephants, while improving the mean read latency by up to 80.4% (25.3% on average) over the state-of-the-art load balancing technique.

AB - Cluster caching has been increasingly deployed in front of cloud storage to improve I/O performance. In shared, multi-tenant environments such as cloud datacenters, cluster caches are constantly contended by many users. Enforcing performance isolation between users hence becomes imperative to cluster caching. A user's caching performance critically depends on two factors: (1) the amount of cache allocation and (2) the load of servers in which its files are cached. However, existing cache sharing policies only provide guarantees on the amount of cache allocation, while remaining agnostic to the load of cache servers. Consequently, 'mice' users having files co-located with 'elephants' contributing heavy data accesses may experience extremely long latency, hence receiving no isolation. In this paper, we propose a Load-Aware Cache Sharing scheme (LACS) to enforce isolation between users. LACS keeps track of the load contributed by each user and reins back the congestions caused by elephant users by throttling their cache usage and network bandwidth. We have implemented LACS atop Alluxio, a popular cluster caching system. EC2 deployment shows that LACS achieves performance isolation in the presence of elephants, while improving the mean read latency by up to 80.4% (25.3% on average) over the state-of-the-art load balancing technique.

KW - Fair cache sharing

KW - Isolation guarantte

UR - http://www.scopus.com/inward/record.url?scp=85074842122&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85074842122&partnerID=8YFLogxK

U2 - 10.1109/ICDCS.2019.00029

DO - 10.1109/ICDCS.2019.00029

M3 - Conference contribution

AN - SCOPUS:85074842122

T3 - Proceedings - International Conference on Distributed Computing Systems

SP - 207

EP - 217

BT - Proceedings - 2019 39th IEEE International Conference on Distributed Computing Systems, ICDCS 2019

PB - Institute of Electrical and Electronics Engineers Inc.

ER -