Towards Virtualization of Rich Applications for Distribution under a

SaaS Model

D. A. Rodríguez-Silva, L. Adkinson-Orellana,V. Fernández-Díaz and F. J. González-Castaño

Gradiant, Network and applications Area, Ed. Citexvi, 36310 Campus, Vigo, Spain

Keywords: Virtualization, Saas, Desktop Applications, Cloud Storage, Remote Desktop, Streaming.

Abstract: Current mobile devices (smartphones, tablets, netbooks…), widely used nowadays, can run potent native

applications, but they cannot support typical desktop applications from any operating system. Since modern

devices support recent web standards as HTML5, it is possible to develop a solution based on a thin web

client to grant remote access to desktop applications offered under a SaaS model. This paper proposes the

development of an innovative remote desktop system able to detect application content and encode it

efficiently in real-time, to support an optimal visualization on clients, combining both remote desktop and

streaming protocols. The system is hosted by a cloud infrastructure that ensures scalability, and it follows a

pay-per-use model. Application providers can include their software in a dynamic cloud repository, from

where it is launched remotely to meet final user demands.

1 INTRODUCTION

Nowadays, the use of mobile devices is increasing to

the point that, according to some forecasts,

smartphones and tablets will displace PCs in the

near future (Want, 2009). Although some mobile

devices do not seem constrained in terms of

computing performance or even display size, the

relationship between computational power and

battery capacity, the input interface or the maximum

number of concurrent applications are also limiting

factors.

Consequently, for some reason or another,

heavy, highly demanding rich applications are not

appropriate for simple user devices, and in most

cases applications functionalities need to be adapted

or reduced to show up in these devices. Cloud-

oriented remote visualization technologies which

allow running an application on a remote high-

performance server and exporting its graphic user

interface to a low-end terminal can be of help. They

also free end-users from the burden of software

maintenance and protection against malicious

attacks.

Remote desktops are well-known visualization

technologies (VNC, RDP, NX…). They usually

perform well with static graphic outputs, as they just

transmit screens differences. However, current

remote desktop technologies present drawbacks, as

the lack of efficient audio support or the poor

visualization of dynamic contents (i.e. video). On

the other hand, even though streaming may offer an

appropriate solution in that case, it results inefficient

when only a small part of the transmitted content is

changing.

In this paper we describe a virtualization cloud

platform that provides optimal remote access to rich

interactive applications through a thin web client,

supported by several recent mobile devices. The

platform relies on a cloud infrastructure that will

also host an application market, following a pay-per-

use business model. Section 2 explains the proposed

architecture and section 3 discusses related work.

Finally, section 4 concludes the paper.

2 SYSTEM ARCHITECTURE

The platform we pursue will allow the execution of

any application in a remote powerful virtual machine

through any client device, without installing any

local software. This will be completely transparent

to the user, regardless of the device capabilities.

The client side will be developed entirely using

standard web technologies, to achieve complete OS

independence. The server side will deploy ready-to-

370

Rodriguez-Silva D., Adkinson-Orellana L., Fernandez-Diaz V. and Gonzalez-Castaño F..

Towards Virtualization of Rich Applications for Distribution under a SaaS Model.

DOI: 10.5220/0004375503700373

In Proceedings of the 3rd International Conference on Cloud Computing and Services Science (CLOSER-2013), pages 370-373

ISBN: 978-989-8565-52-5

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

Figure 1: General architecture of the proposed system.

use applications and will detect motion in screen

content, in order to code any necessary information

intelligently to send it to the client in real-time using

the most appropriate protocol (RFB or streaming) in

each case. To enhance scalability, a cloud

infrastructure will be used. The general architecture

of the system is shown in Figure 1.

2.1 Cloud Infrastructure

The proposed architecture is composed of four

different layers: IaaS, hypervisors, a virtual resource

layer and the SaaS platform. The infrastructure level

provides the upper layers with the physical resources

necessary to run the applications. As the server side

will rely on a virtual machine, a cloud broker is

necessary to manage the cloud infrastructure, in

order to create virtual machines on demand.

As different users share the same virtual

environment, additional security measures are

required to guarantee data isolation. This is achieved

by keeping the user data that the applications access

in a cloud storage instead of in the virtual machine.

How this storage is implemented is open, as it can be

an internal component of the solution or it can be

provided by an external cloud storage service (e.g.

Dropbox). In both cases, the virtualized applications

will manage user files remotely and encryption

techniques will be applied to enhance the cloud

storage mechanism, so user data will remain safe

during the whole process.

Figure 2 shows the cloud architecture, where the

cloud broker is in charge of the virtual resource layer

that provides virtualized applications to the SaaS

platform. Applications will be available through a

repository, in which application providers will be

allowed to upload their software, to be delivered

according to a pay-per-use model. When an

application is uploaded, the repository manager

creates and configures a template for the virtualized

application following the OCCI standard

(http://occi-wg.org). This template is stored in the

application repository, enabling the corresponding

application for user access.

Figure 2: Cloud architecture.

2.2 Remote Visualization

Our virtual machines currently support either

Windows or Linux OS templates, although the

proposed scheme could be extended to other

operating systems.

In addition to the virtualized applications, each

virtual machine includes a RFB server to allow users

to access their applications remotely. The protocol

chosen for RFB communication is VNC (VNC,

1999), one of the most popular ones. This protocol is

conveniently combined with HTTP streaming (see

section 2.3) to improve the user experience at client

side. Each virtual machine can support different

remote desktop user sessions simultaneously through

different displays managed by different VNC servers

in the same machine. Thus, each user working with a

virtualized application will belong to a different OS

session with its own display (Figure 3).

Figure 3: Application virtualization architecture.

In Linux, this application virtualization relies on

Xvfb (http://www.xfree86.org/4.0.1/Xvfb.1.html),

an X window virtual frame buffer, which

Application

Manager

Users

Manager

Hypervisor

IaaS

SaaS Platform

VirtualResource Layer

CloudBroker

Cloud

Storage

Virtual

Applications

Application

Provider

Monitoring &Security

End Users

Repository

Manager

Application

Repository

Cloud applications

Cloud infrastructure

Cloud

storage

Web server

(HTML5 viewer)

Client with

HTML5 browser

Cloud

broker

RFB + streamin

TowardsVirtualizationofRichApplicationsforDistributionunderaSaaSModel

371

implements a virtual X server as a frame buffer in

which different graphical outputs can be dumped,

using several virtual displays, without overlaps. In

Windows there are no virtual graphic devices such

as Xvfb, so each application needs to be virtualized

in a separated virtual machine that contains its own

VNC server.

2.3 Motion Content Detection

The VNC server has three functions:

 Screen delivery of low-motion content

through the VNC protocol.

 HTTP video streaming of high-motion content

 Intelligent switching between VNC and HTTP

streaming protocols depending on the content.

In order to detect whether the screen content is

static or dynamic, the raw images available at the

VNC server are analyzed as follows:

while(true)

fork=0 to N-1 begin

= getFramebuffer(k);

for x = 0 to blocks(fb

)-1

w = 32; h = 32;

k,x

= getBlock(fb

,w,h);

changed =compare(b

k,x

k-1,x

);

if (changed == true)

block_changes

+= 1;

end for

for i=0 to size(block_changes)-1

if (block_changes

> thredshold)

block_changed

= 1;

else block_changed

= 0;

end

end if

detectMotionArea(block_changed);

end for

sleep(t);

end while

The frame buffer content is gathered ‘N’ times in

a loop, grouping the raw pixels in 32×32 blocks. The

blocks are compared with their instances in the

previous iteration and, if at least a pixel differs, the

entire block is marked as a change; on the other

hand, if all the pixels are the same, the block

remains unchanged.

When the comparison loop ends, the number of

changes occurred in each block is checked. If it

exceeds a predefined threshold, the pixels of the

block are set to ‘1’; otherwise, the block is set to ‘0’.

Finally, an array of binary values is obtained,

showing the position of the dynamic blocks in the

screen. Figure 4 shows an example of a VLC player

playing a video inside an Ubuntu desktop and its

corresponding representation as an array with the

motion detected areas marked with ‘1’.

Figure 4: Motion detection process.

Due to the high computational load required to

compare all the pixels of the frame buffer, only a

representative subset is checked. We decided to look

only at those pixels in block diagonals, simplifying

the complexity of the problem from order n

to 2n,

where n is the block width in pixels (assuming a

square block). Nevertheless, the accuracy of the

detection is preserved due to the proximity of the

pixels in a 32×32 block, compared with a typical

1024×720 desktop size. In addition, a hysteresis

criterion is followed to prevent false positives,

improving reliability. Once the motion area is

detected, the VNC server stops sending the RFB

images of that area. Then the stream is generated

and the client receives the URL to play the video in

the corresponding rectangle over the VNC content.

2.4 Web Client

A major concern regarding the visualization of

remote applications is the necessity of a specific

client for each device and platform. To cope with

this we propose the development of a thin client

based on common technologies supported by most

of devices (typically Web over HTTP), following

the latest standards as HTML5. Thus, the web client

may be accessible everywhere through compatible

web browsers.

This client combines and displays contents

received by VNC and HTTP streaming protocols

seamlessly, by using canvas and video tags. Hence,

users will have a web portal containing a catalogue

of the available applications and the client to access

them. The platform has an open-community

orientation, so that users will be able to demand and

CLOSER2013-3rdInternationalConferenceonCloudComputingandServicesScience

372

evaluate applications and software providers will

distribute their applications easily.

3 RELATED WORK

The fact that thin clients improve power efficiency

has been demonstrated previously (Vereecken et al.,

2010). However, a comprehensive study over a wide

range of well-known thin client protocols

(Deboosere et al., 2007) has proved that additional

functionality is required for a satisfying multimedia

experience. In this case streaming is a feasible

mature option, though it may consume lots of

unnecessary bandwidth in case the user interface is

quite static. Several studies have combined remote

desktop and streaming technologies like Simoens,

Praet, Vankeirsbilck, De Watcher and Deboosere

(2008) and Tan, Gong, Wu, Chang and Li (2010).

However, these solutions are not scalable at all, and

they severely limit the number of users that can

access the system. To solve this, some authors have

proposed the use of cloud computing to support

scalable remote visualization through optimized

protocols for mobile devices, although there are still

many open issues (Simoens et al., 2008). In (Zhong

et al., 2010), an approach named vSaas for providing

software as a service from the cloud was described.

Nevertheless, this solution does not offer a smoothly

remote visualization of applications. Shi, Lu, Li and

Engelsma (2010) have present the SHARC solution

for enabling scalable support of real-time 3D

applications in the cloud, delivering content to

clients through a streaming server. However, this

proposal requires installing a specific client or a

Flash browser to visualize the 3D contents.

4 CONCLUSIONS

This paper focuses on the development of a

virtualization platform that will provide optimal

remote access to rich interactive applications

through a thin HTML5 web client supported by

many recent mobile devices. For this to become

possible, it was necessary to apply mechanisms and

algorithms at the server side to automatically detect

changes in the screen content of the applications and

code the dynamic areas in real-time when necessary,

combining optimal protocols to display the screens

at the client side in a efficient way. In addition, the

solution, which is supported by a cloud

infrastructure, includes a repository with

applications that will be delivered as SaaS.

ACKNOWLEDGEMENTS

This research has been supported by Spanish

projects VIMAIN (grant IPT-2011-1224-430000),

funded by MINECO, SECANI (grant

10SEC001CT), funded by Xunta de Galicia, and

partially by CloudMeUp (grant IDI-20101357),

funded by CDTI.

REFERENCES

Deboosere, L., De Wachter, J., Simoens, P., De Turck, F.,

Dhoedt, B., et al. (2007). Thin Client Computing

Solutions in Low- and High-Motion Scenarios. In

Proc. IEEE Third International Conference on

Networking and Services, p. 38-43.

Shi, W., Lu, Y., Li, Z. and Engelsma, J. (2010). Scalable

Support for 3D Graphics Applications in Cloud. In

CLOUD’10, Proc. of IEEE 3rd International

Conference on Cloud Computing.

Simoens, P., Praet, P., Vankeirsbilck, B., De Watcher, J.,

Deboosere, L., De Turk F., et al (2008). Design and

implementation of a hybrid remote display protocol to

optimize multimedia experience on thin client devices.

ATNAC’08, Telecommunication Networks and

Applications Conference.

Tan, K. J., Gong, J. W., Wu, B. T., Chang, D. C., Li, H.

Y., Hsiao Y. M. et al. (2010). A remote thin client

system for real time multimedia streaming over VNC.

In ICME, Proc. of the IEEE International Conference

on Multimedia and Expo.

Vereecken, W., Deboosere, L., Simoens, P., Vermeulen,

B., Colle, D.,Develder, C., et al. (2010). Power

efficiency of thin clients. European Transactions on

Telecommunications, vol. 21, pp. 479-490.

VNC (1999). Virtual Network Computing website.

Retrieved December 4, 2012, from

http://www.hep.phy.cam.ac.uk/vnc_docs/index.html

Want, R. (2009). When cell phones become computers.

IEEE Pervasive Computing, vol. 8, no. 2, pp. 2-5.

Zhong, L., Wo, T., Li, J. and Li, B. (2010). A

Virtualization-Based SaaS Enabling Architecture for

Cloud Computing. In ICAS’10, Proc. of Sixth

International Conference on Autonomic and

Autonomous Systems, Cancun, Mexico.

TowardsVirtualizationofRichApplicationsforDistributionunderaSaaSModel

373