DYNAMIC EVOLUTION OF SERVICE ARCHITECTURE IN

MOBILE CLOUD APPLICATIONS

Huiqun Zhao

, Jing Sun

and Xiaodong Liu

Department of Computer Science, North China University of Technology, Beijing, China

School of Computing, Edinburgh Napier University, Edinburgh, U.K.

Keywords: Service-Oriented Architecture, SaaS in Cloud, Dynamic Evolution, Algebraic Model, Pervasive Systems.

Abstract: Although software services and service-oriented architecture have been researched widely, most existing

research has focused on tools, process and methods of service engineering, and service semantics, and leave

the formal specification of many aspects of SOA unsolved yet. One imminent challenge is the lack of the

fundamental theoretic study of service architecture evolution, which is critical due to the very dynamic

nature of services and service-based systems. This paper contributes to the state of arts by proposing a

methodology that supports dynamic service evolution with respect to an algebra model of SOA. In this

paper we infer and define service evolution as a paradigm of algebraic h

omomorphism

mapping that can

facilitate the analysis of the service architecture evolution with the algebraic method, e.g., the closure and

the consistency of service evolution. In this way we develop two types of different service architecture

evolution: one is for the reconfiguration of service composition inside a service system; and the other is for

the collaboration composition between two service systems. The model has been applied to support the

evolution of cloud-based services (as SaaS) which accept pervasive/mobile accesses. A case study

combining with the transaction management and service adaptation is carried out in a context-aware tourism

service application running in a multi-touch multi-user table providing “intelligent” offering of tourism

information. Conclusions are drawn and future work is identified.

1 INTRODUCTION

Service Oriented Computing is an emerging

paradigm for distributed computing and e-business

processing. The Service Oriented Architecture and

Web Service technique provide a systematic solution

for building an application, for example WSDL,

UDDI, SOAP and BEPL. However, current SOA

technologies and standards are challenged by the

constantly changing environment and user

requirements. Concerns for runtime evolution are

evoked to realize extensible and adaptive service-

oriented applications.

Papazoglou (Papazoglou, 2008) observes the

challenges of service evolution that the service

technologies must be faced to. He classifies four

type of changes in which the service may evolve: 1)

Structural changes, 2) Business protocol changes, 3)

Policy induced changes, and 4) Operational

behavior changes. Mendling and Hafner (Mendling

and Hafner, 2008) propose a novel definition called

service choreography and the service orchestration

for structural changes and protocol changes in

comparison with paper (Papazoglou, 2008), and also

contribute a method to dealing with their evolution.

Andrikopoulos, Benbernou and Papazoglou

(Andrikopoulos et al, 2008) argue that a service

system can provide service evolution management,

which provides an understanding of change impact,

service changes control, tracking and auditing of

service versions, and status accounting. Ryu et al

(Ryu et al, 2008) declare that the business protocol

evolution may raise critical problems: one of the

most critical issues is how to handle instances

running under the old protocol when it has been

changed. Ping Yu, at el (Yu and Ma and Lu, 2005)

introduces a new approach to service evolution.

Taking advantage of experience from software

architecture, especially dynamic software

architecture (DSA) evolution management, they take

service composition into account in runtime

evolution and service evolution is referred to as

dynamic reconfiguration of service architecture. The

method has no formal basic and its viewpoint on

service evolution is different from our algebra model

based approach. Furthermore, many other

271

Zhao H., Sun J. and Liu X..

DYNAMIC EVOLUTION OF SERVICE ARCHITECTURE IN MOBILE CLOUD APPLICATIONS.

DOI: 10.5220/0003899102710274

In Proceedings of the 2nd International Conference on Cloud Computing and Services Science (CLOSER-2012), pages 271-274

ISBN: 978-989-8565-05-1

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

researchers pay much more attention to dynamic

service selection and dynamic service composite

which is related service evolution (Zeng and

Benatallah et al, 2004) (Wan and Ulrich and Chen,

2008)(Zend and Sun, et al, 2010).

In this paper, facing the challenge of lacking the

basic theoretic study of service architecture

evolution, we propose a methodology, referred to as

an algebraic model of SOA, which supports the

evolution design of a dynamic SOA system. We then

define and prove two algebra patterns that can guide

service evolution. The proposed algebraic model has

been applied to support the evolution of cloud-based

services (as SaaS), which accept pervasive/mobile

accesses from clients in multiple modal interactions.

To demonstrate the capability of the algebra

patterns, we use the Apto tool to carry out a case

study, which involves the transaction management

and service adaptation in a context-aware tourism

service application.

The reminder of the paper is organized as

follows: Section 2 introduces the algebraic model of

SOA. Section 3 defines the dynamic evolution of

SOA based on its algebraic model. Section 4

presents the case study of pervasive service

evolution in the cloud. Finally, section 5 presents the

conclusion and future work.

2 AN ALGEBRAIC MODEL OF

SOA

In this section, we define an algebraic model of SOA

based software application. The model provides a

foundation for the software design and the further

evolution of the SOA based application. We

formally define the concepts and verified facts of

SOA in the algebraic model, such as service

component definition, service composition as an

operation and its attributes, the descriptive definition

of SOA.

Definition 2.1 (Service Component). A service

software component (or component for short) comes

from several different forms, including a module of

program code, a programming language, a

computational unit under the execution environment

of the language, a data item, or an item of the data

processing result. A component consists of an

interface and an implementation module.

The interface of a service component is a set P of

external ports, P={Porti, i=1,…,n}. Each port Porti

is an 8-tuple: <ID; Pbuli; Privi; Extni; Behai; Megsi;

Consi; NonFunci>, where

ID is the identifier of the component or a code

for interpreting its IP address.

Publ

is the set of functions (or activities) the

port provides to the external environment and other

components. The execution of an activity is noted as

Extn

is the set of functions (or activities) of the

external environment and other components that the

port interacts with.

Priv

is the set of functions (or activities) used

only internally to this port.

Beha

is a description of the semantics of the

port’s behavior, consisting of a set of predicate logic

expressions.

Megs

is the set of messages generated at this

port, expressed as event signals.

Cons

is the set of constraints to the port

consisting of the initial, pre-and post-conditions for

the execution of the component, and may be

expressed as Cons(init; pre-cond; post-cond). In

general, we can assume that the initial condition is

always satisfied. In these cases, init is omitted in the

expression.

NonFunc

is a description of non-functional

aspects of the port, including descriptions for the

service strategy, contract, security, and reliability.

In the SOA software algebraic model, a set of

operations of service composition are defined,

including “trigger”, “use”, “collaboration”,

“parallel”, “repeat”, and “selection”. Among them,

the operation Trigger depicts the behavior of the

message-based component composition, and is

represented as A ⊕ B or x ⊕ y where x and y are the

activities of component A and B respectively.

The SOA algebraic model clarifies the attributes

of SOA as an algebraic system, and provides the

formal algebraic model definition of SOA. We call

SOA=<C,O> the algebraic model of SOA, also the

algebraic expression of SOA. Here, C and O

represent the set of service components and the set

of the operations of service composition. We here

prove that SOA=<C,O> constructs an algebraic

system, that is:

Theorem 2.1. Let SOA = <C, O>, the SOA

forms an algebraic system to each composition

operation in O.

3 DYNAMIC EVOLUTION OF

SOA STRUCTURE

First we give the definition of the evolution of a

service component.

CLOSER2012-2ndInternationalConferenceonCloudComputingandServicesScience

272

Definition 3.1. Let A and B be two service

components in Dom(U). B is said to be an evolution

from A, denoted Evolve(B;A), if the following

conditions are met:

Dom(B) = Dom(A) (2-2a)

Publ(B) ⊇ Publ(A) (2-2b)

Extn(B) ⊆ Extn(A) (2-2c)

Priv(B) ⊆ Priv(A) (2-2d)

Beha(B) ⇒ Beha(A) (2-2e)

Msgs(B) = Msgs(A) (2-2f)

Cons(B) =Cons(B) or Cons(B)⇒Cons(B)

(2-2g)

NonFunc(B)=NonFunc(A)

or NonFunc(B)⇒NonFunc(A) (2-2h)

The evolution of service component does not

affect the organization of services, therefore has no

direct impact on the architecture, i.e., no change on

the organization of the SOA system. Below we

define the evolution of the architecture and show its

impact on the organization of services in SOA

applications.

Definition 3.2. Let S

=<C

> and S

=<C

> be

two different services. If ∀x∈C

always holds

∃y∈C

and y is a mapping to x, then we assert that

there is a mapping between S

and S

, denoted

f:S

→S

or y=f(x). If the mapping is full mapping,

then the mapping between the services is full

mapping; if the mapping is one to one, the mapping

between the services is also one to one mapping.

Based on definition 3.2, the definition of SOA

architecture evolution is further refined as follows:

Definition 3.3. Let S

=<C

> and S

=<C

> be

two different services. F is a mapping from S

to S

If ∀x∈C

and ∀y∈C

such that f(x⊗y) = f(x) ⊕

f(y),

where ⊗∈O

and

⊕∈O

, then we say the f is

homomorphic evolution from S

to S

. S

and S

are also said to be homomorphic architecture. If f

is a one-way mapping, then f is said to be a one-way

homomorphic evolution from S

to S

. If f is a one-

to-one mapping, then f is said to be an isomorphism

evolution from S

to S

, and S

and S

are

isomorphic architecture.

The evolution of service system may be caused

by various reasons: i) service system evolution

which is caused by the evolution of service

components, which is a passive evolution to the

service developer; ii) service system evolution

which is caused by the evolution of the system

architecture, which is considered as an active

evolution. Active evolution means the improvement

of the quality of services and the original service

system. Here we present one type of the architecture

evolution, which only involves the adjustment of the

mode of service composition (operation).

Theorem 3.1： For any SOA=<C,O>, it is always

possible to evolve into a composition of service

components which only involves the composition of

invocation operations Op^ and selection operations

＋

Theorem 3.2： Let F=a

+…+a

and G=b

+…+b

be two service expressions, then an

isomorphism evolution must definitely exist based

the collaboration operation

，

which makes the

new service have the isomorphism architecture of

ΘG.

4 CASE STUDY

As a proof-of-concept we implemented a prototype

tool (Apto) for the service evolution with process

variant generation technology (Jaroucheh et al,

2010). We present here a case study for a tourism

service application running in a multi-touch multi-

user table that provides “intelligent” offering of

tourism information. The service is deployed as a

mobile SaaS in the cloud “E-Napier”, which consists

of 40 PCs in grid managed by VMWare. The aim of

this application is to provide users with life

experience when booking for their holidays.

Obviously, since usually different users use this type

of applications simultaneously, considering and

resolving conflicts between users’ preferences (part

of the application context) become important. This

application runs in a travel agency which has some

agreements with other travel tourist agents

distributed in different cities. These agents provide

Web service interface for others to get offers and

book for their trip. To give users a life experience

the application displays customized information

about each city i.e. historic, artistic, etc. according to

the users’ preferences. In addition, it tries to find

their friends who are actually located (or have been)

in these cities.

4.1 Specification of the Original System

The original service system architecture is described

in Figure 1. It employs a transactional mode to

serves each user within one event cycle. It is a big

challenge for a decentralized service system to

maintain atomic event cycle especially when the

user number increasing dramatically.

The algebraic model of original service system

present as 4-1.

1( 1 2 3)

2( 2 3)

Service Agent Agent Agent

Servcie Agent Agent

Θ++

⎧

⎨

Θ+

⎩

(4-1)

DYNAMICEVOLUTIONOFSERVICEARCHITECTUREINMOBILECLOUDAPPLICATIONS

273

Figure 1: Original service model.

4.2 Architecture Adaptation

To satisfy the increasing business requirement a

substituted service mode has been adopted which

guided by the Theorem 3.2. For keeping the

consistence with our service model theory, the

formal description of the two service modes is given

as follow.

(11 2 2 3 3)

5( 2 2 3 3)

Service

Service gent Servcie Agent Servcie Agent

Servcie Servcie Agent Servcie Agent

⊗

⎧

⎪

Θ+ Θ + Θ

⎨

⎪

⊗Θ+Θ

⎩

(4-2)

Meanwhile the formula 4-2 lets Service1

collaborates with Agent1-3 individually in according

with the Theorem 3.2. In order to gather the service

information from the parallel Agent1-3, a service

composition, namely service trigger, is executed.

We have designed and implemented a self-

adaptation approach to the service evolution. When

the user number is below a valve we still use the

original service mode, and if the traffic peak

appears, we then change to the new service mode.

5 CONCLUSIONS

The objectives of the research are to develop an

algebraic model for service evolution at both

architecture and service levels and then to identify

and formally define various service evolution

patterns. Our literature review has shown that the

proposed algebraic model has novel contributions

and similar work has not been done.

Based on the algebraic model of SOA, the paper

discusses the methodology of the dynamic evolution

of service architecture, proposes that service

mapping can be viewed as a basis of the evolution of

service architecture, and consequently views

evolution as a process of system transformation

instead of simply focusing on local adjustment. The

paper proposes two fundamental methods for the

dynamic evolution of service architecture, and

demonstrates the feasibility of the methodology

through both theoretical proof and a practical case

study of cloud-based pervasive service evolution.

In the future we will explore the relationships

between the evolution of service components and

service architecture, and the impact of the cloud

features such as dynamic agility and virtualisation

on the architectural evolution of complex cloud-

hosted SOA systems.

ACKNOWLEDGEMENTS

The work in this paper has been jointly sponsored by

the British Royal Society of Edinburgh (RSE-Napier

E4161) and the Natural Science Foundation of China

(Ref: 61070030).

REFERENCES

Andrikopoulos, V., Benbernou, S. and Papazoglou, M. P.,

2008. Managing the Evolution of Service

Specifications. CAiSE 2008: pp. 359-374.

Jaroucheh, Z., Liu, X. and Smith, S. 2010. A MDD-based

Generic Framework for Context-aware Deeply

Adaptive Service-based Processes. The 8th IEEE

International Conference on Web Services (ICWS'10),

Miami, USA.

Mendling, J. and Hafner, M., 2008. From WS-CDL

choreography to BPEL process orchestration, Journal

of Enterprise Information Management, 21(5).

Papazoglou, M.P., 2008. The challenges of Service

Evolution. Advanced Information Systems

Engineering. LNCS, Vol. 5074/2008, pp. 1-15.

Ryu, S.H., Casati, F., Skogsrud, H., Benatallah, B. and

Régis S.P., 2008. Supporting the Dynamic Evolution

of Web Service Protocols in Service- Oriented

Architectures, ACM Transactions on the Web, 2(2).

Wan, C., Ullrich, C., Chen, L. et al., 2008. On solving

QoS aware service selection problem with service

composition. In Proc of the 7

International

Conference on Grid and Cooperative Computing.

Shenzhen: IEEE, pp.467- 474.

Yu, P., Ma, X. and Lu, J., 2005. Dynamic Software

Architecture Oriented Service Composition and

Evolution. Fifth International Conference on

Computer and Information Technology (CIT'05).

Shanghai, China.

Zeng, L. Z., Benatallah, B., Ngu, A. H., Dumas M.,

Kalagnanam, J., and Chang, H., 2004. QoS-aware

middleware for Web services composition”. IEEE

Trans. on Software Engineering, 30(5), pp. 311−327.

Zeng, J., Sun, H., Liu, X., Deng, T., and Huai J., 2010.

Dynamic Evolution Mechanism for Trustworthy

Software Based on Service Composition, Journal of

Software, 21(2), pp.261−276.

Agent

CityC.

Agent



CityB.

Agent



CityA

ServiceEventCycle



ServiceEventCycle



Service

 Service



CLOSER2012-2ndInternationalConferenceonCloudComputingandServicesScience

274