SP-cache: Load-balanced, redundancy-free cluster caching with selective partition

Yinghao Yu, Renfei Huang, Wei Wang, Jun Zhang, Khaled Letaief

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

1 Citation (Scopus)

Abstract

Data-intensive clusters increasingly employ inmemory solutions to improve I/O performance. However, the routinely observed file popularity skew and load imbalance create hot spots, which significantly degrade the benefits of in-memory caching. Common approaches to tame load imbalance include copying multiple replicas of hot files and creating parity chunks using storage codes. Yet, these techniques either suffer from high memory overhead due to cache redundancy or incur nontrivial encoding/decoding complexity. In this paper, we propose an effective approach to achieve load balancing without cache redundancy or encoding/decoding overhead. Our solution, termed SP-Cache, selectively partitions files based on their popularity and evenly caches those partitions across the cluster. We develop an efficient algorithm to determine the optimal number of partitions for a hot file - too few partitions are incapable of mitigating hot spots, while too many are susceptible to stragglers. We implemented SP-Cache in Alluxio, a popular in-memory distributed storage for data-intensive clusters. EC2 deployment and trace-driven simulations show that, compared to the state-of-the-art solution called EC-Cache [1], SP-Cache reduces the file access latency by up to 40% in both the mean and the tail, using 40% less memory.

Original languageEnglish
Title of host publicationProceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1-13
Number of pages13
ISBN (Electronic)9781538683842
DOIs
Publication statusPublished - 11 Mar 2019
Externally publishedYes
Event2018 International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018 - Dallas, United States
Duration: 11 Nov 201816 Nov 2018

Publication series

NameProceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018

Conference

Conference2018 International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018
CountryUnited States
CityDallas
Period11/11/1816/11/18

Fingerprint

Caching
Cache
Redundancy
Partition
Data storage equipment
Decoding
Copying
Hot Spot
Encoding
Resource allocation
Distributed Memory
Replica
Load Balancing
Skew
Parity
Latency
Tail
Efficient Algorithms
Trace
Simulation

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Computer Networks and Communications
  • Hardware and Architecture
  • Theoretical Computer Science

Cite this

Yu, Y., Huang, R., Wang, W., Zhang, J., & Letaief, K. (2019). SP-cache: Load-balanced, redundancy-free cluster caching with selective partition. In Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018 (pp. 1-13). [8665765] (Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/SC.2018.00004

SP-cache : Load-balanced, redundancy-free cluster caching with selective partition. / Yu, Yinghao; Huang, Renfei; Wang, Wei; Zhang, Jun; Letaief, Khaled.

Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018. Institute of Electrical and Electronics Engineers Inc., 2019. p. 1-13 8665765 (Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018).

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

Yu, Y, Huang, R, Wang, W, Zhang, J & Letaief, K 2019, SP-cache: Load-balanced, redundancy-free cluster caching with selective partition. in Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018., 8665765, Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018, Institute of Electrical and Electronics Engineers Inc., pp. 1-13, 2018 International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018, Dallas, United States, 11/11/18. https://doi.org/10.1109/SC.2018.00004
Yu Y, Huang R, Wang W, Zhang J, Letaief K. SP-cache: Load-balanced, redundancy-free cluster caching with selective partition. In Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018. Institute of Electrical and Electronics Engineers Inc. 2019. p. 1-13. 8665765. (Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018). https://doi.org/10.1109/SC.2018.00004
Yu, Yinghao ; Huang, Renfei ; Wang, Wei ; Zhang, Jun ; Letaief, Khaled. / SP-cache : Load-balanced, redundancy-free cluster caching with selective partition. Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 1-13 (Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018).
@inproceedings{1c416992f98b4937b2019845bf566b10,
title = "SP-cache: Load-balanced, redundancy-free cluster caching with selective partition",
abstract = "Data-intensive clusters increasingly employ inmemory solutions to improve I/O performance. However, the routinely observed file popularity skew and load imbalance create hot spots, which significantly degrade the benefits of in-memory caching. Common approaches to tame load imbalance include copying multiple replicas of hot files and creating parity chunks using storage codes. Yet, these techniques either suffer from high memory overhead due to cache redundancy or incur nontrivial encoding/decoding complexity. In this paper, we propose an effective approach to achieve load balancing without cache redundancy or encoding/decoding overhead. Our solution, termed SP-Cache, selectively partitions files based on their popularity and evenly caches those partitions across the cluster. We develop an efficient algorithm to determine the optimal number of partitions for a hot file - too few partitions are incapable of mitigating hot spots, while too many are susceptible to stragglers. We implemented SP-Cache in Alluxio, a popular in-memory distributed storage for data-intensive clusters. EC2 deployment and trace-driven simulations show that, compared to the state-of-the-art solution called EC-Cache [1], SP-Cache reduces the file access latency by up to 40{\%} in both the mean and the tail, using 40{\%} less memory.",
author = "Yinghao Yu and Renfei Huang and Wei Wang and Jun Zhang and Khaled Letaief",
year = "2019",
month = "3",
day = "11",
doi = "10.1109/SC.2018.00004",
language = "English",
series = "Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "1--13",
booktitle = "Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018",

}

TY - GEN

T1 - SP-cache

T2 - Load-balanced, redundancy-free cluster caching with selective partition

AU - Yu, Yinghao

AU - Huang, Renfei

AU - Wang, Wei

AU - Zhang, Jun

AU - Letaief, Khaled

PY - 2019/3/11

Y1 - 2019/3/11

N2 - Data-intensive clusters increasingly employ inmemory solutions to improve I/O performance. However, the routinely observed file popularity skew and load imbalance create hot spots, which significantly degrade the benefits of in-memory caching. Common approaches to tame load imbalance include copying multiple replicas of hot files and creating parity chunks using storage codes. Yet, these techniques either suffer from high memory overhead due to cache redundancy or incur nontrivial encoding/decoding complexity. In this paper, we propose an effective approach to achieve load balancing without cache redundancy or encoding/decoding overhead. Our solution, termed SP-Cache, selectively partitions files based on their popularity and evenly caches those partitions across the cluster. We develop an efficient algorithm to determine the optimal number of partitions for a hot file - too few partitions are incapable of mitigating hot spots, while too many are susceptible to stragglers. We implemented SP-Cache in Alluxio, a popular in-memory distributed storage for data-intensive clusters. EC2 deployment and trace-driven simulations show that, compared to the state-of-the-art solution called EC-Cache [1], SP-Cache reduces the file access latency by up to 40% in both the mean and the tail, using 40% less memory.

AB - Data-intensive clusters increasingly employ inmemory solutions to improve I/O performance. However, the routinely observed file popularity skew and load imbalance create hot spots, which significantly degrade the benefits of in-memory caching. Common approaches to tame load imbalance include copying multiple replicas of hot files and creating parity chunks using storage codes. Yet, these techniques either suffer from high memory overhead due to cache redundancy or incur nontrivial encoding/decoding complexity. In this paper, we propose an effective approach to achieve load balancing without cache redundancy or encoding/decoding overhead. Our solution, termed SP-Cache, selectively partitions files based on their popularity and evenly caches those partitions across the cluster. We develop an efficient algorithm to determine the optimal number of partitions for a hot file - too few partitions are incapable of mitigating hot spots, while too many are susceptible to stragglers. We implemented SP-Cache in Alluxio, a popular in-memory distributed storage for data-intensive clusters. EC2 deployment and trace-driven simulations show that, compared to the state-of-the-art solution called EC-Cache [1], SP-Cache reduces the file access latency by up to 40% in both the mean and the tail, using 40% less memory.

UR - http://www.scopus.com/inward/record.url?scp=85064109338&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85064109338&partnerID=8YFLogxK

U2 - 10.1109/SC.2018.00004

DO - 10.1109/SC.2018.00004

M3 - Conference contribution

AN - SCOPUS:85064109338

T3 - Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018

SP - 1

EP - 13

BT - Proceedings - International Conference for High Performance Computing, Networking, Storage, and Analysis, SC 2018

PB - Institute of Electrical and Electronics Engineers Inc.

ER -