Advanced

Efficient Provisioning of Bursty Scientific Workloads on the Cloud Using Adaptive Elasticity Control

Ali-Eldin, Ahmed; Kihl, Maria LU ; Tordsson, Johan and Elmroth, Erik (2012) ScienceCloud 3rd Workshop on Scientific Cloud Computing In [Host publication title missing]
Abstract
Elasticity is the ability of a cloud infrastructure to dynamically

change the amount of resources allocated to a running

service as load changes. We build an autonomous elasticity

controller that changes the number of virtual machines allocated

to a service based on both monitored load changes

and predictions of future load. The cloud infrastructure is

modeled as a G=G=N queue. This model is used to construct

a hybrid reactive-adaptive controller that quickly reacts

to sudden load changes, prevents premature release of

resources, takes into account the heterogeneity of the workload,

and avoids oscillations. Using simulations with Web

and... (More)
Elasticity is the ability of a cloud infrastructure to dynamically

change the amount of resources allocated to a running

service as load changes. We build an autonomous elasticity

controller that changes the number of virtual machines allocated

to a service based on both monitored load changes

and predictions of future load. The cloud infrastructure is

modeled as a G=G=N queue. This model is used to construct

a hybrid reactive-adaptive controller that quickly reacts

to sudden load changes, prevents premature release of

resources, takes into account the heterogeneity of the workload,

and avoids oscillations. Using simulations with Web

and cluster workload traces, we show that our proposed controller

lowers the number of delayed requests by a factor of

70 for the Web traces and 3 for the cluster traces when compared

to a reactive controller. Our controller also decreases

the average number of queued requests by a factor of 3 for

both traces, and reduces oscillations by a factor of 7 for

the Web traces and 3 for the cluster traces. This comes at

the expense of between 20% and 30% over-provisioning, as

compared to a few percent for the reactive controller. (Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
in
[Host publication title missing]
publisher
ACM
conference name
ScienceCloud 3rd Workshop on Scientific Cloud Computing
external identifiers
  • Scopus:84863924453
project
LCCC
language
English
LU publication?
yes
id
49e6a5f6-a4e2-44cc-b6c5-b098ebc53e35 (old id 2539762)
date added to LUP
2012-05-15 10:08:09
date last changed
2017-02-12 04:24:08
@inproceedings{49e6a5f6-a4e2-44cc-b6c5-b098ebc53e35,
  abstract     = {Elasticity is the ability of a cloud infrastructure to dynamically<br/><br>
change the amount of resources allocated to a running<br/><br>
service as load changes. We build an autonomous elasticity<br/><br>
controller that changes the number of virtual machines allocated<br/><br>
to a service based on both monitored load changes<br/><br>
and predictions of future load. The cloud infrastructure is<br/><br>
modeled as a G=G=N queue. This model is used to construct<br/><br>
a hybrid reactive-adaptive controller that quickly reacts<br/><br>
to sudden load changes, prevents premature release of<br/><br>
resources, takes into account the heterogeneity of the workload,<br/><br>
and avoids oscillations. Using simulations with Web<br/><br>
and cluster workload traces, we show that our proposed controller<br/><br>
lowers the number of delayed requests by a factor of<br/><br>
70 for the Web traces and 3 for the cluster traces when compared<br/><br>
to a reactive controller. Our controller also decreases<br/><br>
the average number of queued requests by a factor of 3 for<br/><br>
both traces, and reduces oscillations by a factor of 7 for<br/><br>
the Web traces and 3 for the cluster traces. This comes at<br/><br>
the expense of between 20% and 30% over-provisioning, as<br/><br>
compared to a few percent for the reactive controller.},
  author       = {Ali-Eldin, Ahmed and Kihl, Maria and Tordsson, Johan and Elmroth, Erik},
  booktitle    = {[Host publication title missing]},
  language     = {eng},
  publisher    = {ACM},
  title        = {Efficient Provisioning of Bursty Scientific Workloads on the Cloud Using Adaptive Elasticity Control},
  year         = {2012},
}