A forecasting methodology for workload forecasting in cloud systems

Francisco J. Baldan, Sergio Ramirez-Gallego, Christoph Bergmeir, Francisco Herrera, Jose M. Benitez-Sanchez

Research output: Contribution to journalArticleResearchpeer-review

Abstract

IEEE Cloud Computing is an essential paradigm of computing services based on the “elasticity” property, where available resources are adapted efficiently to different workloads over time. In elastic platforms, the forecasting component can be considered by far the most important element and the differentiating factor when comparing such systems, being workload forecasting one of the problems to solve if we want to achieve a truly elastic system. When properly addressed the cloud workload forecasting problem becomes a really interesting case study. As there is no general methodology in the literature that addresses this problem analytically and from a time series forecasting perspective (even less in the cloud field), we propose a combination of these tools based on a stateof- the-art forecasting methodology which we have enhanced with some elements, such as: a specific cost function, statistical tests, visual analysis, etc. The insights obtained from this analysis are used to detect the asymmetrical nature of the forecasting problem and to find the best forecasting from the viewpoint of the current state of the art in time series forecasting. From an operational point of view the most interesting forecast is a short-time horizon, so we focus on this. To show the feasibility of this methodology, we apply it to several realistic workload datasets from different datacenters. The results show that the analyzed series are non-linear in nature, and that no seasonal patterns can be found. Moreover, on the analyzed datasets, the penalty cost as usually included in the SLA can be reduced down to a 30% on average.
Original languageEnglish
Pages (from-to)929-941
Number of pages13
JournalIEEE Transactions on Cloud Computing
Volume6
Issue number4
DOIs
Publication statusPublished - 1 Dec 2018

Keywords

  • Business
  • Cloud computing
  • Cloud Computing
  • Computational modeling
  • Elasticity
  • elasticity
  • Forecasting
  • machine learning
  • Predictive models
  • Time series analysis
  • time series forecasting
  • workload forecasting

Cite this

Baldan, Francisco J. ; Ramirez-Gallego, Sergio ; Bergmeir, Christoph ; Herrera, Francisco ; Benitez-Sanchez, Jose M. / A forecasting methodology for workload forecasting in cloud systems. In: IEEE Transactions on Cloud Computing. 2018 ; Vol. 6, No. 4. pp. 929-941.
@article{08ca71923c8b423d95caab4a277e2fa9,
title = "A forecasting methodology for workload forecasting in cloud systems",
abstract = "IEEE Cloud Computing is an essential paradigm of computing services based on the “elasticity” property, where available resources are adapted efficiently to different workloads over time. In elastic platforms, the forecasting component can be considered by far the most important element and the differentiating factor when comparing such systems, being workload forecasting one of the problems to solve if we want to achieve a truly elastic system. When properly addressed the cloud workload forecasting problem becomes a really interesting case study. As there is no general methodology in the literature that addresses this problem analytically and from a time series forecasting perspective (even less in the cloud field), we propose a combination of these tools based on a stateof- the-art forecasting methodology which we have enhanced with some elements, such as: a specific cost function, statistical tests, visual analysis, etc. The insights obtained from this analysis are used to detect the asymmetrical nature of the forecasting problem and to find the best forecasting from the viewpoint of the current state of the art in time series forecasting. From an operational point of view the most interesting forecast is a short-time horizon, so we focus on this. To show the feasibility of this methodology, we apply it to several realistic workload datasets from different datacenters. The results show that the analyzed series are non-linear in nature, and that no seasonal patterns can be found. Moreover, on the analyzed datasets, the penalty cost as usually included in the SLA can be reduced down to a 30{\%} on average.",
keywords = "Business, Cloud computing, Cloud Computing, Computational modeling, Elasticity, elasticity, Forecasting, machine learning, Predictive models, Time series analysis, time series forecasting, workload forecasting",
author = "Baldan, {Francisco J.} and Sergio Ramirez-Gallego and Christoph Bergmeir and Francisco Herrera and Benitez-Sanchez, {Jose M.}",
year = "2018",
month = "12",
day = "1",
doi = "10.1109/TCC.2016.2586064",
language = "English",
volume = "6",
pages = "929--941",
journal = "IEEE Transactions on Cloud Computing",
issn = "2168-7161",
publisher = "IEEE, Institute of Electrical and Electronics Engineers",
number = "4",

}

Baldan, FJ, Ramirez-Gallego, S, Bergmeir, C, Herrera, F & Benitez-Sanchez, JM 2018, 'A forecasting methodology for workload forecasting in cloud systems' IEEE Transactions on Cloud Computing, vol. 6, no. 4, pp. 929-941. https://doi.org/10.1109/TCC.2016.2586064

A forecasting methodology for workload forecasting in cloud systems. / Baldan, Francisco J.; Ramirez-Gallego, Sergio; Bergmeir, Christoph; Herrera, Francisco; Benitez-Sanchez, Jose M.

In: IEEE Transactions on Cloud Computing, Vol. 6, No. 4, 01.12.2018, p. 929-941.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - A forecasting methodology for workload forecasting in cloud systems

AU - Baldan, Francisco J.

AU - Ramirez-Gallego, Sergio

AU - Bergmeir, Christoph

AU - Herrera, Francisco

AU - Benitez-Sanchez, Jose M.

PY - 2018/12/1

Y1 - 2018/12/1

N2 - IEEE Cloud Computing is an essential paradigm of computing services based on the “elasticity” property, where available resources are adapted efficiently to different workloads over time. In elastic platforms, the forecasting component can be considered by far the most important element and the differentiating factor when comparing such systems, being workload forecasting one of the problems to solve if we want to achieve a truly elastic system. When properly addressed the cloud workload forecasting problem becomes a really interesting case study. As there is no general methodology in the literature that addresses this problem analytically and from a time series forecasting perspective (even less in the cloud field), we propose a combination of these tools based on a stateof- the-art forecasting methodology which we have enhanced with some elements, such as: a specific cost function, statistical tests, visual analysis, etc. The insights obtained from this analysis are used to detect the asymmetrical nature of the forecasting problem and to find the best forecasting from the viewpoint of the current state of the art in time series forecasting. From an operational point of view the most interesting forecast is a short-time horizon, so we focus on this. To show the feasibility of this methodology, we apply it to several realistic workload datasets from different datacenters. The results show that the analyzed series are non-linear in nature, and that no seasonal patterns can be found. Moreover, on the analyzed datasets, the penalty cost as usually included in the SLA can be reduced down to a 30% on average.

AB - IEEE Cloud Computing is an essential paradigm of computing services based on the “elasticity” property, where available resources are adapted efficiently to different workloads over time. In elastic platforms, the forecasting component can be considered by far the most important element and the differentiating factor when comparing such systems, being workload forecasting one of the problems to solve if we want to achieve a truly elastic system. When properly addressed the cloud workload forecasting problem becomes a really interesting case study. As there is no general methodology in the literature that addresses this problem analytically and from a time series forecasting perspective (even less in the cloud field), we propose a combination of these tools based on a stateof- the-art forecasting methodology which we have enhanced with some elements, such as: a specific cost function, statistical tests, visual analysis, etc. The insights obtained from this analysis are used to detect the asymmetrical nature of the forecasting problem and to find the best forecasting from the viewpoint of the current state of the art in time series forecasting. From an operational point of view the most interesting forecast is a short-time horizon, so we focus on this. To show the feasibility of this methodology, we apply it to several realistic workload datasets from different datacenters. The results show that the analyzed series are non-linear in nature, and that no seasonal patterns can be found. Moreover, on the analyzed datasets, the penalty cost as usually included in the SLA can be reduced down to a 30% on average.

KW - Business

KW - Cloud computing

KW - Cloud Computing

KW - Computational modeling

KW - Elasticity

KW - elasticity

KW - Forecasting

KW - machine learning

KW - Predictive models

KW - Time series analysis

KW - time series forecasting

KW - workload forecasting

UR - http://www.scopus.com/inward/record.url?scp=85028540439&partnerID=8YFLogxK

U2 - 10.1109/TCC.2016.2586064

DO - 10.1109/TCC.2016.2586064

M3 - Article

VL - 6

SP - 929

EP - 941

JO - IEEE Transactions on Cloud Computing

JF - IEEE Transactions on Cloud Computing

SN - 2168-7161

IS - 4

ER -