P-store

An elastic database system with predictive provisioning

Rebecca Taft, Nosayba El-Sayed, Marco Serafini, Yu Lu, Ashraf Aboulnaga, Michael Stonebraker, Ricardo Mayerhofer, Francisco Andrade

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

3 Citations (Scopus)

Abstract

OLTP database systems are a critical part of the operation of many enterprises. Such systems are often configured statically with sufficient capacity for peak load. For many OLTP applications, however, the maximum load is an order of magnitude larger than the minimum, and load varies in a repeating daily pattern. It is thus prudent to allocate computing resources dynamically to match demand. One can allocate resources reactively after a load increase is detected, but this places additional burden on the already-overloaded system to reconfigure. A predictive allocation, in advance of load increases, is clearly preferable. We present P-Store, the first elastic OLTP DBMS to use prediction, and apply it to the workload of B2W Digital (B2W), a large online retailer. Our study shows that P-Store outperforms a reactive system on B2W's workload by causing 72% fewer latency violations, and achieves performance comparable to static allocation for peak demand while using 50% fewer servers.

Original languageEnglish
Title of host publicationSIGMOD 2018 - Proceedings of the 2018 International Conference on Management of Data
PublisherAssociation for Computing Machinery
Pages205-219
Number of pages15
ISBN (Electronic)9781450317436
DOIs
Publication statusPublished - 27 May 2018
Event44th ACM SIGMOD International Conference on Management of Data, SIGMOD 2018 - Houston, United States
Duration: 10 Jun 201815 Jun 2018

Other

Other44th ACM SIGMOD International Conference on Management of Data, SIGMOD 2018
CountryUnited States
CityHouston
Period10/6/1815/6/18

Fingerprint

Servers
Industry

ASJC Scopus subject areas

  • Software
  • Information Systems

Cite this

Taft, R., El-Sayed, N., Serafini, M., Lu, Y., Aboulnaga, A., Stonebraker, M., ... Andrade, F. (2018). P-store: An elastic database system with predictive provisioning. In SIGMOD 2018 - Proceedings of the 2018 International Conference on Management of Data (pp. 205-219). Association for Computing Machinery. https://doi.org/10.1145/3183713.3190650

P-store : An elastic database system with predictive provisioning. / Taft, Rebecca; El-Sayed, Nosayba; Serafini, Marco; Lu, Yu; Aboulnaga, Ashraf; Stonebraker, Michael; Mayerhofer, Ricardo; Andrade, Francisco.

SIGMOD 2018 - Proceedings of the 2018 International Conference on Management of Data. Association for Computing Machinery, 2018. p. 205-219.

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

Taft, R, El-Sayed, N, Serafini, M, Lu, Y, Aboulnaga, A, Stonebraker, M, Mayerhofer, R & Andrade, F 2018, P-store: An elastic database system with predictive provisioning. in SIGMOD 2018 - Proceedings of the 2018 International Conference on Management of Data. Association for Computing Machinery, pp. 205-219, 44th ACM SIGMOD International Conference on Management of Data, SIGMOD 2018, Houston, United States, 10/6/18. https://doi.org/10.1145/3183713.3190650
Taft R, El-Sayed N, Serafini M, Lu Y, Aboulnaga A, Stonebraker M et al. P-store: An elastic database system with predictive provisioning. In SIGMOD 2018 - Proceedings of the 2018 International Conference on Management of Data. Association for Computing Machinery. 2018. p. 205-219 https://doi.org/10.1145/3183713.3190650
Taft, Rebecca ; El-Sayed, Nosayba ; Serafini, Marco ; Lu, Yu ; Aboulnaga, Ashraf ; Stonebraker, Michael ; Mayerhofer, Ricardo ; Andrade, Francisco. / P-store : An elastic database system with predictive provisioning. SIGMOD 2018 - Proceedings of the 2018 International Conference on Management of Data. Association for Computing Machinery, 2018. pp. 205-219
@inproceedings{33b0567bf4954b9cae46dba2f2cc69e3,
title = "P-store: An elastic database system with predictive provisioning",
abstract = "OLTP database systems are a critical part of the operation of many enterprises. Such systems are often configured statically with sufficient capacity for peak load. For many OLTP applications, however, the maximum load is an order of magnitude larger than the minimum, and load varies in a repeating daily pattern. It is thus prudent to allocate computing resources dynamically to match demand. One can allocate resources reactively after a load increase is detected, but this places additional burden on the already-overloaded system to reconfigure. A predictive allocation, in advance of load increases, is clearly preferable. We present P-Store, the first elastic OLTP DBMS to use prediction, and apply it to the workload of B2W Digital (B2W), a large online retailer. Our study shows that P-Store outperforms a reactive system on B2W's workload by causing 72{\%} fewer latency violations, and achieves performance comparable to static allocation for peak demand while using 50{\%} fewer servers.",
author = "Rebecca Taft and Nosayba El-Sayed and Marco Serafini and Yu Lu and Ashraf Aboulnaga and Michael Stonebraker and Ricardo Mayerhofer and Francisco Andrade",
year = "2018",
month = "5",
day = "27",
doi = "10.1145/3183713.3190650",
language = "English",
pages = "205--219",
booktitle = "SIGMOD 2018 - Proceedings of the 2018 International Conference on Management of Data",
publisher = "Association for Computing Machinery",

}

TY - GEN

T1 - P-store

T2 - An elastic database system with predictive provisioning

AU - Taft, Rebecca

AU - El-Sayed, Nosayba

AU - Serafini, Marco

AU - Lu, Yu

AU - Aboulnaga, Ashraf

AU - Stonebraker, Michael

AU - Mayerhofer, Ricardo

AU - Andrade, Francisco

PY - 2018/5/27

Y1 - 2018/5/27

N2 - OLTP database systems are a critical part of the operation of many enterprises. Such systems are often configured statically with sufficient capacity for peak load. For many OLTP applications, however, the maximum load is an order of magnitude larger than the minimum, and load varies in a repeating daily pattern. It is thus prudent to allocate computing resources dynamically to match demand. One can allocate resources reactively after a load increase is detected, but this places additional burden on the already-overloaded system to reconfigure. A predictive allocation, in advance of load increases, is clearly preferable. We present P-Store, the first elastic OLTP DBMS to use prediction, and apply it to the workload of B2W Digital (B2W), a large online retailer. Our study shows that P-Store outperforms a reactive system on B2W's workload by causing 72% fewer latency violations, and achieves performance comparable to static allocation for peak demand while using 50% fewer servers.

AB - OLTP database systems are a critical part of the operation of many enterprises. Such systems are often configured statically with sufficient capacity for peak load. For many OLTP applications, however, the maximum load is an order of magnitude larger than the minimum, and load varies in a repeating daily pattern. It is thus prudent to allocate computing resources dynamically to match demand. One can allocate resources reactively after a load increase is detected, but this places additional burden on the already-overloaded system to reconfigure. A predictive allocation, in advance of load increases, is clearly preferable. We present P-Store, the first elastic OLTP DBMS to use prediction, and apply it to the workload of B2W Digital (B2W), a large online retailer. Our study shows that P-Store outperforms a reactive system on B2W's workload by causing 72% fewer latency violations, and achieves performance comparable to static allocation for peak demand while using 50% fewer servers.

UR - http://www.scopus.com/inward/record.url?scp=85048821650&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85048821650&partnerID=8YFLogxK

U2 - 10.1145/3183713.3190650

DO - 10.1145/3183713.3190650

M3 - Conference contribution

SP - 205

EP - 219

BT - SIGMOD 2018 - Proceedings of the 2018 International Conference on Management of Data

PB - Association for Computing Machinery

ER -