Cloud-Based Multimedia Storage System with QoS Provision

Swati Vani et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 5 (2) , 2014, 1173-1176
Cloud-Based Multimedia Storage System with QoS
Swati Vani, Bhagyashri Bhosale, Ganesh Shinde, Rajni Shinde, Prof.Manoj Pawar.
Department of Information Technology
Marathwada Mitra Mandal’s Institute of Technology, Lohgaon, Pune-411043, India
Abstract—Recently there are new trends in the way we use
computers and access networks due to advanced mobile devices
and network technologies. One of trend is cloud computing
where resources are stored and processed on network. The other
is Mobile computing, where mobile devices such as smart phones
and tablets combines network connectivity, mobility, and
software functionality and working as personal computers.
Cloud based multimedia services have high constraint in terms of
bandwidth and jitter. Therefore different approaches required to
manage resources more efficiently for better Quality of Service
(QoS) and Quality of Experience (QoE) offered by the mobile
media services. This paper introduces a novel concept of Mobile
Multimedia Web Service using Cloud in which services will run
on public cloud depending upon service demands and network
status, the service will be populated on other public cloud in
different geographical locations. If demand for particular service
increases in a location it will be more reliable to populate that
service[1] closer to the cloud in that location. This will prevent
the high traffic loads on internet backbone due to streaming of
multimedia data. It will offer service provider’s management
mechanism and an automated resource allocation for their
services. This will help to reduce bandwidth and jitter on the
cloud based multimedia services.
IndexTerms—Computer Network Management, Communication
System, Web Services, Mobile Computing.
Amazon Machine Images (AMI) and several possible virtual
machines, which differ in CPU power, memory and disk
space. This functionality allows freely select suitable
technologies for any particular task. In case of Amazon EC2
price for the service depends on machine size, its up time, and
used bandwidth in and out of the cloud.
Cloud computing is one of the trends in IT which refers to
Compute Cloud (EC2): is a central part of
application and services that run on distributed network using’s
computing platform, Amazon Web
virtualized resources and accessing by common Internet
users to rent virtual computers
protocols and networking standards. There are three categories
applications. EC2 allows
of cloud services: Software as a Service (SaaS), Platform as a
Service (PaaS), and Infrastructure as a Service (IaaS). SaaS scalable deployment of applications by providing a Web
delivers software application over the internet. Google Apps service through which a user can boot an Amazon Machine
(includes Google Mail, Docs, Sites, Calendar, etc) is the Image to create a virtual machine, which Amazon calls
example of SaaS. PaaS delivers a host operating system and an ’INSTANCE’, containing any software desired.
development tools which installed virtualized resources. The iCloud[2]: The service allows users to store data such as
example of PaaS is Google App Engine which provides elastic music and iOS applications on remote computer servers for
platform for Java and Python applications. IaaS offers number download to multiple devices such as iOS-based devices
of virtual machines or processors and storage space and leaves running iOS. The service also allows users to wirelessly back
it up to the user to select how these resources are used. up their iOS devices to iCloud instead of manually doing so
Amazon EC2[3] (example of IaaS) are probably most known using iTunes.
and widely used. Amazon EC2 provides an instance of a Office 365: It is a subscription-based online office and softvirtual machine image that allows full control over the ware plus services suite which offers access to various
operating system. It is possible to select a suitable operating services and software built around the Microsoft Office
system, and platform (32 and 64 bit) from many available platform.
Swati Vani et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 5 (2) , 2014, 1173-1176
Cloud based service Layered Framework: We relate the layers
of the architecture with the OSI model. The proposed
framework and the OSI model share the same level of
abstraction in terms of network technologies and protocols
and this makes it easy to use the OSI as a reference to our
model as opposed to using the TCP/IP model.
The service architecture is not meant to map directly to some
of the OSI layers. Some of the functions performed in the
proposed layers can interact with OSI layers to perform
network-level operations while other layers do not present any
functions that directly interface with the OSI and are therefore
considered extra layers.
The Service Management Layer (SML): Deals with how
services are registered in a Cloud. This also includes the
overall Service and Security Level Agreement (SSLA)
between the Cloud providers and the service providers and the
unique Service ID. The SML can be considered as part of the
Application Layer in the OSI since it defines the applications
themselves and how they use resources.
The Service Subscription Layer (SSL): Deals with the
subscription of clients to the service and holds information
that handles the subscriptions such as User IDs, the list of
services subscribed to by individual client and the associated
client SLAs between clients and services. This layer can give
instructions to the Presentation Layer in the OSI in order to
handle user specific service parameters such as encryption or
CODECs in video streams. The SSL can be considered as part
of the Application Layer in the OSI.
The Service Delivery Layer (SDL): Is responsible for the
delivery of services to individual clients. The layers below
receive instructions from this layer with regard to connecting
to individual clients as well as populating Clouds.
The Service Migration Layer (SMiL): Is responsible for the
Migration of services between Clouds. It deals with resource
allocation across Clouds to facilitate service population. It also
holds the mechanism that performs the handover of client
connections between services. The SSL can be considered as
part of the Application Layer in the OSI.
The Service Connection Layer (SCL): Monitors connections
between clients and services. Some of this layer’s functions
map directly to the Session Layer in the OSI model.
Service Network Abstraction Layer (SNAL): Makes the
network technology transparent to the upper layers in order to
simplify and unify the process of migration. The function of
this layer is to act as a common interface between the service
delivery framework and the underlying network architecture
such as IP overlay network [6] or new technologies which
divide the Internet into a Core network surrounded by
Peripheral wireless networks.
Abstraction of service layer: In SML when a service
provider wishes to publish a service, they have to define
security and QoS parameters [4]. In SDL, the logic that
processes all the data regarding QoS characteristics and user
mobility resides in this layer. It uses data from the overall
SSLA and the client SLA and checks if the requirements are
met by using network QoS data given by the layer below.
Such data can be fed to this layer by the mobile devices
themselves either in the form of a process running separately
or through a QoS-aware protocol that can report latency and
bandwidth between two end points. The Cloud that fulfills all
the parameters in the SSLA list and can provide better QoS
than the others can then proceed to the Migration process in
the layer below. In SCL the SCL is also responsible for the
network handover between clients and services after a service
moves. This is done by gathering QoS data from the network
and from client devices.
Implementation mechanism: In order to gather QoS data and
know the network conditions in a specific area, we are using
another mechanism that we call the QoS Monitor. It is
considered to be part of the SCL and acquires such data by
querying the clients for network conditions. The mechanism
that we are assuming here that can resolve human-friendly
service names to unique Service IDs. In the SDL we need
mechanisms that will connect service subscribers to the
correct instance of a service for service delivery purposes. A
record of Service IDs and in which Clouds their instances are
running and also uses input by the QoS Tracking are
maintained by the Service Tracking and Resolution or STAR.
STAR will make a decision on which Cloud is better suited to
service a client request based on the location of the client,
using this information.
STAR achieve this functionality is by look up routing tables in
order to identify which Cloud is closer to a user. Service
delivery mechanism using STAR is shown in fig.2 Service to
reject the new client and forward them to another Cloud if
possible. This gives control to service providers and also
becomes a contingency mechanism in case STAR makes a
wrong decision. The STAR server can be scaled similarly to
the DNS[5] system since it is essentially the same type of
Swati Vani et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 5 (2) , 2014, 1173-1176
service albeit with some extra parameters. Once a Cloud ID is displayed and the hardware capabilities of the mobile device.
found, then the ID is resolved into the IP addresses of the We now substitute for Tprefetch in (3) with the expressions in
Cloud controllers that the client can contact to access the (1) and (2). Rearranging, we get:
service. The process is shown in the Fig. 3. It should be noted Fn,s,θ+Fcloud+FProtocol≥(Tcpu-C)*p
that alternatively the Cloud ID can be returned to the client, at
which point, the client will have a choice of which DNS to use Exploring network latency in detail, for each link we have
to find the IP addresses.
transmission delay and queuing delay. Therefore, the total
network latency will be the sum of the latencies for each link
between client and service. Hence, we can express as:
If we denote the transport block size as b, then the time to
transmit p blocks over a link is equal to the number of blocks
multiplied by the block size and divided by the bandwidth of
the link.
So, we have,
Fcloud+FProtocol+∑((p*b)/Si+Qi)≥ (Tcpu-C)*p
On a lightly loaded system, we consider Fprotocol, Fcloud and Qi
to be negligible.
∑(b/Si) )≥ (Tcpu-C)
Let be the soft limit that we are aiming for in order to prevent
jitter and SL is the migration time.
Where al is the rate of network latency increase as the number
We start by defining the time to prefetch blocks of data, which of network links increases. We can calculate al at the mobile
is given by:
device and we can also find Mt between two Clouds. HL is
given by the mobile device, so we can calculate to SL find
where to set out QoS trigger for service migration.
In this equation, L is the network latency and C is the per We can visualized how the increasing number of links
block time of copying data between the in-cache memory and between a user and a service can bring the connection near the
network buffers. Ideally should be at least equal to the number QoS limit and how we can use a soft limit to trigger service
of blocks required to display a video frame of data. On a migration in order to prevent this. We can also see that for a
lightly loaded wired network we can consider these values given migration time, we need to adjust SL so that during the
constant for each link. However, in a mobile environment, migration the QoS will not reach the HL.
changes as the client moves and the number of network links
increase. We can express L as follows:
From a computational perspective, Cloud providers can share
their resources with other providers. This gives them the
Where, (Fn,s,θ) is the latency incurred by the number of links(n) flexibility to request additional resource when their Cloud
between client and service, the network bandwidth on each needs them or rent some of their resources to other providers
link (Si) and the network load on each link (θi), Fcloud is the that need them.
Cloud latency caused by the network topology and hierarchy By taking into account multimedia creation services such as
within the Cloud Fprotocol is the latency caused by the transport rendering, we can see how such a scenario is applicable and
how it can benefit clients and providers alike. Furthermore, if
If the time to prefetch blocks is larger than the time it takes for we combine the above scenario with mobile devices, we can
the device to consume them, then we have jitter. This can be see how in the future we may find ourselves in a position
where rendering is done on the Cloud and the mobile devices
expressed as:
only display the content.
This can occur in applications such as games. In these
Where (Tcpu) the time it takes for a device to consume a situations, the proposed framework will not only balance the
number of blocks by playing them as audio and video frames. rendering load on Clouds but will also relieve networks from
(Tcpu) is therefore dependent on the type of video being the high traffic generated by streaming video and audio. The
Swati Vani et al, / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 5 (2) , 2014, 1173-1176
distance reduction between clients and services caused by
migrations will also decrease the latency and give users a
more interactive feel to their multimedia application, thus
improving the QoE.
In this paper, we discuss the challenges which are faced by the
mobile user in future networks. The service delivery models
which are used currently are not that much sufficient and not
consider the needs of mobile user in future.
A cloud storage system was proposed in order to provide
robust, scalable, highly available and load-balanced services.
In the meantime, the system also needs to provide quality of
service provision for multimedia applications and services.
The proposed system achieves the three functions of a
multimedia-aware cloud: 1) QoS supporting and provisioning,
2) Parallel processing in distributed environment, 3) QoS
adaptation. These functions make the proposed system
especially suitable to the video on demand service. it often
provides different service quality to users with various types
of devices and network bandwidth.
We believe that our implementation will provide the better
quality of service (QoS) as well as better quality of experience
(QoE) to the user.
Fragkiskos Sardis, Glenford Mapp, Jonathan Loo, Mahdi Aiash, Member, IEEE, and Alexey Vinel, Senior Member, IEEE On the
Investigation of Cloud-Based Mobile Media Environments With
Apple, 2012. iCloud Feb. 15, 2012. [Online]. Available: http://www.
Amazon, 2012, EC2, Feb. 28, 2012. [Online]. Available: http://aws.
D. Gupta, S. Lee, M. Vrable, S. Savage, A. C. Snoeren, G. Varghese, G.
M. Voelker, and A. Vahdat, Difference engine: Harnessing memory
redundancy in virtual machines,? in Proc. OSDI, 2008.
W. Zhu, C. Luo, J. Wang, and S. Li, Multimedia cloud computing, IEEE
Signal Process. Mag., vol. 28, no. 3, pp. 5969, May 2011.
T. Brisko, RFC 1794, DNS Support for Load Balancing, IETF, 1995.
Y-Comm Research, Middlesex Univ., Mar. 2, 2012. [Online]. Available: search:aspx: