Optimal Allocation of Virtual Machines for
Cloud-based Multimedia Applications
Xiaoming Nan, Yifeng He, Ling Guan
Department of Electrical and Computer Engineering
Ryerson University, Toronto, Ontario, Canada
Abstract—With the emergence of cloud computing, cloud-
based multimedia applications have been increasingly adopted
in recent years. There are two major challenges for multimedia
application providers: the round trip time (RTT) requirement
and the resource cost. In this paper, we study the virtual machine
(VM) allocation problem for multimedia application providers to
minimize the resource cost under RTT requirements. Specifically,
we propose the optimal VM allocation schemes for single-site
cloud and multi-site cloud, respectively. Moreover, we propose
the greedy algorithms to efficiently allocate VMs in each case.
Simulation results demonstrate that the proposed optimal VM
allocation schemes can optimally allocate VMs to achieve a
minimal resource cost.
I. INTRODUCTION
As the emerging computing paradigm, cloud computing
manages a shared pool of servers in data centers to provide on-
demand computing resources (e.g. computing power, platform,
applications, etc.) as accessible services for users via the
Internet. To enable the resource provisioning efficiently, the
virtualization technology is applied to package the required
CPU, memory, and storage resources into virtual machines
(VMs). By using VMs, cloud resources can be provisioned or
released with the minimal efforts. According to the service
provisioning at different levels, three cloud service models
have been proposed [1], namely Infrastructure as a Service
(IaaS), Platform as a Service (PaaS), and Software as a Service
(SaaS), among which the SaaS model is the most familiar to
individual cloud users.
In the SaaS cloud model, three major entities are in-
volved in the service provisioning: cloud providers, applica-
tion providers, and users. As the computing resource suppliers,
cloud providers operate the infrastructure of data centers and
provide VMs as service. By renting VMs from cloud provider-
s, application providers can deliver different applications to
users. As the consumers in the SaaS model, users send requests
for interested applications and receive results from application
providers. All these three entities can benefit from the SaaS
model. Users don’t need to install applications or buy software
licences, application providers don’t have to make investment
on servers, and cloud providers can acquire profits from VMs
rental.
Among various cloud-based applications, multimedia appli-
cations strongly need assistance from cloud computing due
to the compute-intensive and delay-sensitive requirements.
For multimedia application providers, there are two major
challenges: the round trip time (RTT) and the resource cost.
The RTT in cloud is defined as the sum of the forward trans-
mission delay from the user to the data center, the backward
transmission delay from the data center to the user, and the
service response time in the data center. Considering the delay-
sensitive requirement of multimedia, the RTT is taken as a
significant Quality of Service (QoS) factor to measure the
performance of cloud-based multimedia applications. But it
is a challenge to satisfy the RTT requirements of different
multimedia applications for users in different regions. Besides
the RTT, the resource cost is also a challenge. Generally, cloud
providers can offer two VM pricing schemes: the reservation
scheme and the on-demand scheme. Price of VMs in the reser-
vation scheme is cheaper than that in the on-demand scheme.
But in the reservation scheme, application providers have to
subscribe a certain amount of VMs in advance for future usage.
During the service provisioning, if the reserved VMs cannot
meet the resource demands, the on-demand VMs can be rent
instantly to meet the extra demands at the expense of a higher
price rate. Facing different applications and resource demands,
it is challenging for application providers to optimally allocate
VMs for each application to achieve the minimal resource cost
and satisfy the different RTT requirements.
In this paper, we investigate the VM allocation problem
for multimedia application providers to minimize the resource
cost under the RTT requirements. Our contributions in this
paper can be presented as follows. We propose the optimal VM
allocation schemes for single-site cloud and multi-site cloud,
respectively. In each case, the theoretical optimal solution
and practical greedy algorithm are proposed for multimedia
application providers. Specifically, we formulate and solve the
optimal VM allocation problem to minimize the resource cost
under the RTT constraints. Since the formulated optimization
problem is an NP-hard problem, we also propose the greedy
algorithm to efficiently allocate VMs. The greedy algorithm
is a sub-optimal solution, which is demonstrated to perform
close to the optimal solution by the simulation results.
II. RELATED WORK
Cloud computing, as the emerging computing paradigm, has
widely attracted attentions of researchers from both industry
[2], [3], [4] and academia [5-14]. Survey works on the general
definition, challenges, and service models of cloud computing
can be found from [5], [6]. Among different cloud service
models, SaaS model has been identified as the most profitable
model. According to the Gartner Group estimate [7], the
,((( 0063