Position Paper: Multi-tenants Context-aware Service Composition in

Cloud Computing

Wael Sellami, Hatem Hadj Kacem and Ahmed Hadj Kacem

ReDCAD Lab, University of Sfax, Sfax, Tunisia

Keywords:

Cloud Service Composition, Context-awareness, Multi-tenancy, Adaptation.

Abstract:

Cloud computing has gained a momentum by providing multi-tenancy utility and applications as a service.

The power of services occurs in the ability to combine different services in order to obtain a composed ser-

vice. Context-aware cloud service is anticipated to bring an innovation in mobile computing. Therefore the

composed service can be a multi-tenant context-aware, scalable and adaptable. In this paper, we deﬁne the

problem of coupling context-aware cloud service composition with multi-tenancy in the cloud and we deter-

mine crucial challenges that can be solved. For that, we propose an approach to enable multiple providers to

run their composed services within the same instance, with different contexts and without modiﬁcations of the

workﬂows.

1 INTRODUCTION

Cloud computing emerged as one of the most power-

ful and ﬂexible software environment. Brieﬂy, NIST

(Mell and Grance, 2011) deﬁnes cloud computing as

a promising and challenging model which delegates

the material’s management and software applications

which users pay as they go. Services in cloud com-

puting are the combination of utility computing ser-

vices and service oriented architecture in order to

provide resources as a service classiﬁed into three

layers: SaaS (Software as a Service), PaaS (Palte-

form as a Service) or IaaS (Infrastructure as a Ser-

vice). In fact, the PaaS layer is a model consisting

of elements capable of managing business processes

through tools and technologies such as multi-tenancy,

scalability/elasticity, data persistence and son on. The

multi-tenancy is a key feature in PaaS layer. It means

that only one instance of each service is shared by all

tenants through virtualization technologies and isola-

tion. With the advent of mobile cloud computing and

the proliferation of the customized services, the prin-

ciple of multi-tenancy can be inﬂuenced by context-

awareness applications. According to (Dey, 2001),

the context includes information that can character-

ize the situation of a user’s local environment and

the computing devices themselves. These applica-

tions should take into account the situations of ten-

ants context change or applications which may inﬂu-

ence their behavior by deﬁning new views on their

data and services. Cloud services composition is de-

ﬁned as the process of linking existing software and

hardware services. However, it is different from SOA

and utility computing in terms of web or grid service

form. Therefore, it is hard to use traditional service

matching methods by the characteristics of dynami-

cally conﬁgured, massively scalable, distributed re-

sources and multi-tenancy architecture of cloud ser-

vices. Indeed, combining the concepts of context-

aware cloud service composition with multi-tenancy

is complex and causes new problems. Indeed, ap-

plying each technique independently and serially pro-

vides a unique view of the composed cloud service

for all tenants regardless of their preferences and con-

texts. Furthermore, adaptation of the composition

during sharing between all tenants.

In this position paper, we aim to develop a new

middleware for establishing the context-aware cloud

service matching and adaptation. Indeed, this middle-

ware must be able to carry out a process of adaptation

in order to deliver the most suitable composed service

to tenants not only according to their queries but also

in relation to their contexts for ensuring a high perfor-

mance and respecting SLA clauses.

This paper is organized as follows. Section 2 in-

troduces our motivation while Section 3 gives a back-

ground of multi-tenant architecture on the cloud. In

Section 4, we present the related works. Section 5 dis-

cusses our initial approach and Section 6 deals with

conclusions and ongoing works.

386

Sellami W., Hadj Kacem H. and Hadj Kacem A..

Position Paper: Multi-tenants Context-aware Service Composition in Cloud Computing.

DOI: 10.5220/0004406303860389

In Proceedings of the 3rd International Conference on Cloud Computing and Services Science (CLOSER-2013), pages 386-389

ISBN: 978-989-8565-52-5

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

2 MOTIVATION

Many service providers are interested in using the

multi-tenancy, scalability and other beneﬁts of the

cloud (Paraiso et al., 2012). With the rapid develop-

ment of mobile cloud computing, context-aware in-

formation (such as personal, environmental and op-

erational contexts) can satisfy tenants with their re-

ally needed composed services without any explicit

requests. However, conﬂicts can be generated by as-

sembling the multi-tenancy with the service compo-

sition context-aware. Indeed, there are antagonistic

concepts since the aim of multi-tenancy is the abil-

ity to use the same application to multiple tenants.

While the purpose of the context-aware composition

is to suit a single application for multiple tenants with

different contexts and preferences. To understand the

problem well, we can take two scenarios of multi-

tenancy and context-aware service composition:

A. Multi-tenancy

• Scenario. Executing a composed cloud service

for products research into on-line market.

• Result. Sharing the composed services instance

to provide all images and descriptions of all the

products sorted out by category for all tenants.

B. Context-aware Service Composition

• Scenario. Executing the same composed cloud

services for products research into on-line market

from a mobile phone.

• Result. Providing product description only with

women’s preferences.

Towards the two scenarios, the instance of composed

cloud services has the same behavior and all the used

data are shared between all tenants regardless of spe-

ciﬁc contexts and requirements for each tenant. So,

this challenge leads us to ask a question:

How can we realize a multi-behavior instance of a

composed service for all tenants ?

Combining the concept of multi-tenancy with

context-aware cloud service composition to establish

new compromise middlware is complex and provides

new and interesting settings for both issues.

3 A CONCEPT FOR MULTI

TENANCY ARCHITECTURE

In cloud computing, multi-tenant architecture is the

ability of hosting multiple users under a single server

and sharing the same infrastructure and applications

(Domingo et al., 2010). A multi-tenant application

must support multiple tenants which can have mul-

tiple users and user group preferences. In order to

meet the different tenant requirements and the vari-

ation context, services must be highly customizable

and conﬁgurable. So, service providers have to offer

certain level of variability in their multi-tenants appli-

cation, and they should be aware of the level of the

variability, in terms of its complexity impact on ser-

vice ﬂexibilities. From our perspective, we think that

the balance between composed cloud services and the

variability of context has to be well maintained. This

can be done by integrating service facilities on a per

tenant (as it is shown in Figure 1) and modifying the

system behavior to suit the tenant responses.

Figure 1: Variation points of multi-tenant composite ser-

vice.

4 RELATED WORKS

The problem of cloud service matching and adapta-

tion has received a lot of attention in academia and

industry in the recent years. In this section, we present

in the ﬁrst part service composition approaches and in

the second part we give an overview of services adap-

tation.

4.1 Cloud Service Composition

A new approach for services composition is presented

in (Tsai et al., 2011). This work is based on the mech-

anism of service selection using service interfaces and

service injection techniques. Thus, it proposes a se-

ries of tests to check services and ensure the reliability

of composition. In (Ye et al., 2011), authors propose a

multi-cloud service composition approach. Users re-

quirement is translated into data ﬂow graphs. Then,

by using a genetic algorithm, a service composition

solution is provided. Furthermore, in (Koﬂer et al.,

2010), authors determine a heuristic based approach

PositionPaper:Multi-tenantsContext-awareServiceCompositioninCloudComputing

387

to compose services in cloud computing. They focus

on user’s functional and non-functional requirements.

Thus, an algorithm is applied to automate the selec-

tion of optimal services and to realize service com-

position Another approach to build a decentralized

framework based on self organized agents is deter-

mined in (Garcia and Sim, 2010). Authors used con-

tract net protocol for dynamic communications tech-

nical knowledge of acquintance networks for incom-

plete information about existing services. In (Zeng

et al., 2009), authors solve the problem of cloud ser-

vice composition using a service matching algorithm

(SMA). It is based on QoS to ﬁlter services and per-

form the matching algorithm of services based on the

semantic similarity between input and output param-

eters to achieve this composition.

We notice that these works do not provide appropriate

support for the concept of context and progress sce-

narios during cloud services searching and matching.

Moreover, they do not retain the concept of the cloud

which is the multi-tenancy nature during composition

where a single instance is shared between tenants.

4.2 Cloud Service Adaptation

The approach presented in (Truyen et al., 2012) de-

scribes a framework for tenant-speciﬁc customization

and manages changes of software-tenant SaaS appli-

cations. To achieve this, this framework is based on

the context oriented programming to provide ﬂexible

SaaS services for multi-tenants suitable to the context

information. In (Papakos et al., 2010), authors de-

velop a middleware called VOLARE which monitors

resources and context of mobile devices, and dynam-

ically suits the cloud services. This approach allows a

reliable services discovery and cost reduction at run-

time to ﬁt in the current context of the consumers. In

(La and Kim, 2010), authors propose a framework for

context-awareness services for mobile computing. In

fact, this framework allows the capture of the con-

text, determines the speciﬁc adaptation and executes

the right services.

However, these efforts focus only on the customiza-

tion of a single service but no solution has been pro-

posed to resolve the adaptation of composed cloud

service. Also, they deal with the personalization of

responses at a high level but not the personalization

of services at a lower level before the composition to

provide more ﬂexibility of applications.

To overcome these issues, we propose in this paper a

primal tenant cloud service composition middleware

to address all aspects of the context awareness starting

from the dynamic services matching until to adapta-

tion.

5 INITIAL APPROACH

In this section we present, in detail, our multi-tenants

context-aware cloud service composition middleware.

Indeed (as it is shown in Figure 2), we aim to focus on

two steps of the life cycle: (1) services matching and

(2) adaptation of a composed service. Each step con-

tains components, and their related communications.

5.1 Tenant Users

Tenant users layer is composed of multiple tenant

applications deployed on portable multimedia player

and mobile device such as Android or SmartPhone.

Thus, we consider that a tenant can be a single user

or a whole company consisting of a collection of dis-

tinct users which can access to his ﬂexible services

whenever he wants and wherever he is.

5.2 Service Matching

A composition engine is a multi-tenant service ex-

ecuting on the cloud platform. It parses the tenant

query and uses the contexts information. Then, it

can be able to lookup its cloud services from repos-

itories component using a given policy (such as the

nearest data center to tenant location, the fastest re-

sponse if the tenant is running, and so on) and choose

the most appropriate composition that suits the ten-

ant’s context. In fact, once the services are selected,

the composition engine is responsible for deﬁning a

sequence of services to be executed and supervising

the composed service. The composition engine can

decide which concrete services are included in the dy-

namic composition, and how many instances of each

concrete service are needed in order to satisfy the ten-

ant’s context. Therefore, this component can dynam-

ically update the composition and substitute any con-

crete service at any time with another service based

on the feedback of the tenant’s runtime context data.

5.3 Service Adaptation

In context-aware composite service provisioning,

adaptation can be achieved by splitting it into sepa-

rate services and tailoring them for the given context.

Tenant speciﬁc customization can be speciﬁed as in-

dependent software variations according to some cat-

egories of context such as tenant proﬁle, environment

information and so on. For that, we propose the run-

time behavior adaptation component which is called

tenant context adaptor in order to suit the composi-

tion, by retrieving context information from the Con-

text Base registry, and updating it to new situations.

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

388

Figure 2: Multi-tenants context-aware cloud service composition middleware.

This adaptation is used to integrate new behaviors and

concerns as persistence or security into the original

cloud service without affecting other applications that

share the same service with other tenants. After that,

the multi-instance balancer component distributes the

instance for each tenant according to his global con-

text and to satisfy his requirements.

6 CONCLUSIONS AND

ONGOING WORK

Due to the open and ﬂexible nature of cloud ser-

vices, there is a gap between service composition ef-

ﬁciency and the tenants ﬂexible response needs. In

this paper, a context-aware cloud service composi-

tion middleware is investigated for improving the ser-

vices matching efﬁciency in mobile cloud environ-

ment. Concretely, this middleware permits dynamic

context-aware composition and adaptation of ﬂexi-

ble cloud based services, and how these character-

istics inﬂuence the multi-tenancy adaptation. In the

ongoing work, we are developing a middleware and

are deﬁning the experimental techniques to simulate

cloud service composition according to the context-

aware. Thus, we are aiming to exploit new compo-

sition scenarios in a real environment to analyze the

middleware efﬁciency. For future work, we will detail

each component of our middlware and propose efﬁ-

cient techniques for implementing them. Moreover,

we will enhance the adaptation of the tenant compos-

ite service with other contextual situations.

REFERENCES

Dey, A. (2001). Understanding and using context. Journal

of Pervasive and Ubiquitous Computing.

Domingo, E., Nino, J., Lemos, A., Lemos, M., Palacios, R.,

and Berbs, J. (2010). Cloudio: A cloud computing-

oriented multi-tenant architecture for business infor-

mation systems. IEEE 3rd International Conference

on Cloud Computing.

Garcia, J. and Sim, K. (2010). Self-organizing agents for

service composition in cloud computing. Interna-

tional Conference on Cloud Computing Technology

and Science.

Koﬂer, K., Haq, I., and Schikuta, E. (2010). User-

centric, heuristic optimization of service composition

in clouds. International Euro-Par conference on Par-

allel processing.

La, H. and Kim, S. (2010). A conceptual framework for

provisioning context-aware mobile cloud services. In-

ternational Conference on Cloud Computing.

Mell, P. and Grance, T. (2011). The nist deﬁnition of cloud

computing.

Papakos, P., Capra, L., and Rosenblum, D. S. (2010).

Volare: Context-aware adaptive cloud service discov-

ery for mobile systems. 9th Workshop on Adaptive

and reﬂective Middleware.

Paraiso, F., Haderer, N., Merle, P., Rouvoy, R., and Sein-

turier, L. (2012). A federatedmulti-cloud paas infras-

tructure. IEEE 5th International Conference on Cloud

Computing (CLOUD).

Truyen, E., Cardozo, N., Walraven, S., Vallejos, J., Baino-

mugisha, E., Gunther, S., D’Hondt, T., and Joosen, W.

(2012). Context-oriented programming for customiz-

able saas applications. 27th Annual ACM Symposium

on Applied Computing.

Tsai, W., Zhong, P., Balasooriya, J., Chen, Y., Bai, X., and

Elston, J. (2011). An approach for service composi-

tion and testing for cloud computing. International

Symposium on Autonomous Decentralized Systems.

Ye, Z., Zhou, X., and Bouguettaya, A. (2011). Genetic algo-

rithm based qos-aware service compositions in cloud

computing. International conference on Database

systems for advanced applications.

Zeng, C., Guo, X., Ou, W., and Han, D. (2009). Cloud

computing service composition and search based on

semantic. International Conference on Cloud Com-

puting.

PositionPaper:Multi-tenantsContext-awareServiceCompositioninCloudComputing

389