A STUDY ON ENTERPRISE APPLICATION INTEGRATION

BASED ON GRID

Dan Chang and Lijun Zhou

Institute of Information Systems, Beijing Jiaotong University, Beijing, China

Key words: Grid Technology, Enterprise Application Integration, EAI-Based-Grid.

Abstract: Enterprise Application Integration (EAI) is a kind of method and technique which can solve the problem of

information sharing. However, the existing EAI solutions generally have some limitations. For example,

when the enterprises need to implement cross-boundary information integration between different systems,

they cannot meet the needs. The grid technology can achieve cross-boundary, cross-platform and distributed

resource sharing, also provide complete security mechanism. So the enterprise application integration based

on grid technology is able to solve the current problems. In this paper, some existing studies on EAI

solutions will be introduced. And then grid technology will be applied to the enterprise application

integration to form a solution EAI-Based-Grid, at last we analyze the characteristic of this method and

conduct some simulations to prove its feasibility.

1 INTRODUCTION

With the rapid development of e-commerce,

enterprises have adopted ERP, CRM, PDM and

other information systems successively which

greatly improve the level of information technology

application. However, these systems may be

implemented by different companies and only focus

on their respective data and business processes.

Because of the lack of corresponding standard

specifications for interface, they cannot be carried

out information sharing and business integrations

each other. Therefore, companies need an increasing

strong and interconnecting IT infrastructure to face

to new opportunities and challenges. Enterprise

Application Integration (Qureshi, 2005) (EAI) is a

kind of method and technique which solves the

matter of information exchange and sharing. It uses

an open, extensible system to integrate enterprise

software and hardware, finally achieve the resource

sharing and business processes.

The present EAI solutions (Hub-And-Spoke,

message bus, web services (Wu and Yang, 2008),

etc.) may partially meet the enterprises’ demands for

information sharing between systems, while there

are limitations of the application, such as the

platform restrictions, geographical restrictions or

safety visits. Especially when companies need to

achieve cross-boundary information integration

between different systems, this limitation is

particularly prominent. However, the grid

technology can implement cross-boundary,

cross-platform and distributed resource sharing, also

provide complete security mechanism (Zhang and

Bao, 2008). The enterprise application integration

based on grid technology can solve the current

problems.

In this paper, we propose the advantages and

disadvantages of existing EAI solution. Directing at

their limitations and combining with the advantages

of grid technology, we propose enterprise

application integration solution based on grid

through absorbing the essence of Reference 4. And

then we conduct simulation about this method. The

simulation results prove the method that grid

technology is applied to enterprise application

integration is feasible.

The second part in this paper briefly describes

the existing studies of enterprise applications

integration. The third part explains the theories and

knowledge about grid technology. According to

these two parts, enterprise application integration

based on grid is proposed in the forth part. Then in

order to prove the feasibility of this method, a

simulation is carried out to demonstrate it. At the

end of this paper, it is a general summary.

292

Zhou L. and Chang D.

A STUDY ON ENTERPRISE APPLICATION INTEGRATION BASED ON GRID.

DOI: 10.5220/0003268802920298

In Proceedings of the Twelfth International Conference on Informatics and Semiotics in Organisations (ICISO 2010), page

ISBN: 978-989-8425-26-3

2 EXISTING STUDIES ON EAI

SOLUTIONS

The existing EAI solutions include Hub-And-Spoke,

message bus, web services, etc. All of them have

their own advantages and disadvantages.

2.1 Hub-and-Spoke

In the Hub-And-Spoke structure, different

application programs are connected to a central

server. As is shown in Figure 1 (Gu et al., 2004), a

variety of relational databases, customer relationship

management systems and so on are connected to a

central server through the adapter. The server is like

a hub, it is similar as the star structure of network

topology. Central server plays the role of a message

broker and it is mainly responsible for the

communication management, data transmission and

process between multiple systems. A new system or

application program can be connected to other hub

through an adapter.

Figure 1: Hub-And-Spoke structure.

Hub-And-Spoke structure possesses some

advantages such as centralized control, ease of

administration and so on. However the disadvantage

is that the server easily becomes a bottleneck in the

system. The whole system’s stability will be

gradually decreased when the new system is

connected.

2.2 Message Bus Structure

In the message bus structure, all systems are

connected to the virtual message bus via adapter.

The adapter will convert the internal events of

application program into message and then exchange

the message with other application programs via

message bus. Meanwhile it will receive the message

and convert the message into internal events. It is

shown in Figure 2 (Gu et al., 2004).

Figure 2: Message bus structure.

The advantage of message bus structure is that it

has no communication bottleneck, the adding of new

system is easy to do and scalability is wonderful. On

the contrary the drawback is that the implementation

is complex.

2.3 Web Service

The web services structure can integrate enterprise

applications (Wu and Yang, 2008). The discovery,

description and transmission of services are run by

web services based on UDDI, WSDL and SOAP

agreement of XML. This system reduces the

coupling of enterprise application integration. It is

shown in Figure 3.

Web services structure has the advantage of low

coupling, good cross-platform and a wide range of

application. However, security, transaction and

workflow are its weak point.

A STUDY ON ENTERPRISE APPLICATION INTEGRATION BASED ON GRID

293

Figure 3: Web service structure.

Figure 4: Hourglass model.

3 RELATED THEORIES ON

GRID

3.1 Grid

Grid is originally used for large-scale scientific and

engineering computing, and which now has

developed into an important technology for

distributed resource sharing. The integration of

remote computing resources can be carried out by

grid technology. Then the dynamic resources sharing

and problem processing of multi-agency virtual

organization will be realized 。 The problem

mentioned here may involve data processing, data

storage or business processing (Iosup et al., 2006).

A system can be named grid must meet the

following three points (Foster et al., 2001): i)

Sharing the non-centralized and synergic resource.

ii) Using standard and open protocols and interfaces.

iii) Providing wonderful quality of service.

3.2 Standard and Architecture of Grid

The international organization which draws up grid

computing standards, Global Grid Forum, has

released standard system called OGSA (Benson et

al., 2006) (Open Grid Service Architecture) and

OGSI (Wang et al., 2004) (Open Grid Service

Interface), which provides the structure and interface

of grid respectively.

The major problem of distributed resource

sharing is interoperability. In the network

environment, interoperability depends on the

common agreement. Therefore, the grid architecture

involves the structure of related agreement first, and

then the working mechanism is built on the

agreement structure. The design of grid architecture

follows the Hourglass Model principles (Zhang and

Bao, 2008). It is shown in Figure 4.

In the grid environment, there are a large number

of temporary services, such as a computing task

execution. Considering the specific characteristics of

grid environment, OGSA propose grid service

ICISO 2010 - International Conference on Informatics and Semiotics in Organisations

294



Figure 5: Framework of EAI-Based-Grid.

based on original web services. It is used to solve

the

issues which are related with temporary service

such as service discovery, dynamic service creation,

and service life-cycle management.

4 EAI BASED ON GRID

TECHNOLOGY

4.1 Structure of EAI-based-Grid

Considering the demerits of the previous three kinds

of EAI structure and the merits of grid, EAI based

on grid technology can be proposed. EAI-based-Grid

adopts OGSA as basic framework. In OGSA,

everything, including computers, programs, data and

equipment so on, is abstracted as service. Globus

Toolkit software package and Web Service are two

supporting technologies for the construction of

OGSA. Globus Toolkit is a set of services and

software library that supports grid computing and

grid application. Web Service is commonly used as a

standard framework to access network applications.

The framework of EAI-Based-Grid is shown in

Figure 5.

After entering grid portal through web browser,

the client can see the unified data view of system,

access and manipulate the data, and call the grid

service provided by system. The communication

between the client and the grid server is carried out

by a grid service interface. The transfer protocol

between them is the SOAP protocol and the data is

described by XML language.

In this structure, programming language of grid

service interface can use the high-level service

programming language provided by Globus Toolkit.

This interface offers methods of the access to all

internal grid resources. Every service node has an

operation platform of service, such as Apache, J2EE,

WebSphere and so on, that supports Grid Service

and conducts interaction with the upper layer (Grid

Service Management and Discovery layer).

4.2 Characteristic of EAI-based-Grid

(i) Grid portal simplifies the process of access to

resources. It is convenient for user to interact with

multiple systems.

The client can find all the resources in the system

through the grid portal, including grid services,

operational status, technical documentation, etc. The

users who have a system account and be authorized

can add and manipulate their files and data, also can

submit, monitor and abolish the interaction with the

relevant application system. In this way, users can

achieve seamless access to system resources through

web browser. However, the other enterprise

application integration is difficult to achieve this

result.

(ii) The application interface layer is intergraded

through grid service interface, which improve the

intelligent level of the application system.

When a user requests grid services through a grid

portal, grid interface convert the user’s request to

A STUDY ON ENTERPRISE APPLICATION INTEGRATION BASED ON GRID

295

Figure 6: Process of simulation model one.

Figure 7: Process of simulation model two.

XML files, send it to the task analysis module and

then form a macro-task workflow. For every stage of

the workflow, analysis module will query whether

corresponding grid services are included through the

grid service management and discovery layer. If

these services are found, the task will be optimized

first and then a grid services list related to the task

will be made. Next these services will be called and

the results of the service will be saved in Cache.

Finally it will be submit to the client through the

grid portal. If these services are not found, an error

returns. In this way, the dynamic integration of

application interfaces is realized and the intelligent

level of the system is improved.

(iii) Grid service management and discovery layer

uses the popular UDDI technical specifications,

which is convenient for integration with existing

systems.

It takes use of service registry technology and UDDI

standard to complete grid service registration of each

server node. If the grid needs to perform specific

calculation, the grid service management and

discovery layer will register the service to node that

owns such service and activate the service.

(iiii) Server nodes based on grid can provide a

variety of grid services and improve comprehensive

utilization of various resources in the system.

These services include hardware status services,

message services, basic business function library

services and application services. These services can

ICISO 2010 - International Conference on Informatics and Semiotics in Organisations

296

be aware by grid service management and discovery

layer after they are registered, then be managed and

distributed by grid service interface layer and finally

be provided to clients by grid portal. Clients can take

use of all the resources of the system in accordance

with their permissions. It improves the rate of

multipurpose utilization dramatically.

5 SIMULATION AND ANALYSIS

To verify the advantages of EAI-Based-Grid

mentioned above, we need make the solution analog

abstracted and then we can carry out simulation and

analysis on it. And in order that simulations between

based on non-grid and grid have comparability, we

assume that the resources and the parameters in the

two systems (non-grid and grid) are the same.

5.1 Simulation Model

Model One: Simulation model based on Non-grid.

In this model, if the user requests service X, he need

to query n (i=1, 2… n) resources of each server in

order to find the node that can meet the demand.

Resource i has m links such as occupation, delay,

release and so on. All resources subject to the FIFO

rule. If all resources are occupied, it will form a

queue. The client can not get the service until the

resource is idle. The process of simulation model is

shown as Figure 6.

Model Two: Simulation model based on Grid. In

accordance with the grid management theory, this

model integrates business process to form a unified

control center. Then enterprise business integration

and resource sharing will be realized.

In this model, if the user requests service X, the

service can be aware by grid service management

and discovery layer after register. Then the service is

managed and distributed by grid service interface

layer. Finally, the service is provided to client

through the grid portal. If the resource X is

occupied, a single queue will be formed.

If the

resource X is idle, the system will arrange it for the

user. The process of simulation model is shown as

Figure 7.

5.2 Simulation Results

Based on the above model, the results of simulation

using Arena software are shown as Figure 8, Figure

9, and Figure 10. The horizontal axis represents

system resources, the vertical axis respectively

shows three different indexes, Average Length of

Queue, Average Sojourn Time, Level of Resources

Utilization.

Comparison about Average Length of

Queue

0.01

0.02

0.03

0.04

0.05

0 20 40 60 80 100

System Resources

I:Non-grid II:Grid

Figure 8: Comparison about Average Length of Queue.

The Average Length of Queue L:



∫

∞→

dttSt

))()((

lim

[S (t) represents customers

who are receiving service.

()

t represents

customers who are waiting]. The simulation result of

comparison about average length of queue is shown

as Figure 8. As our reasoning in the Characteristic of

EAI-Based-Grid, the average length of queue in

non-grid pattern is longer than that in grid pattern.

Comparison about Average Sojourn Time

0.01

0.02

0.03

0.04

0.05

0 20406080100

Syst em Resources

I:Non-grid II:Grid

Figure 9: Comparison about Average Sojourn Time.

The Average Sojourn Time W:



∑

∞→

SiZi

)(

lim

[Si represents time of receiving

service. Zi represents time of waiting].The

simulation result of comparison about average

sojourn time is shown as Figure 9. As our reasoning

in the Characteristic of EAI-Based-Grid, the average

sojourn time in non-grid pattern is longer than that in

grid pattern.

A STUDY ON ENTERPRISE APPLICATION INTEGRATION BASED ON GRID

297

Comparison about Level of Resources

Ut ilizat ion

0.2

0.4

0.6

0.8

020406080100

Sy s t e m Re so ur c es

I:Grid II:Non-grid

Figure 10: Comparison about Level of Resources

Utilization.

The Level of resources utilization u:

()

lim

tdt

→∞

∫

[B(t) Represents resources

running status, busy to take 1,idle to take 0]. The

simulation result of comparison about level of

resources utilization is shown as Figure 10. As our

reasoning in the Characteristic of EAI-Based-Grid,

the level of resources utilization in grid pattern is

higher than that in non-grid pattern.

From the above results, the pattern based on grid

is better than that based on non-grid in the three

major indicators .Therefore, the idea of using grid

technology in enterprise application integration is

feasible in a sense.

6 CONCLUSIONS

Considering the limitations of existing EAI solutions

and the advantages of grid technology, we propose

enterprise application integration based on grid

technology. EAI-Based-Grid solves the problem of

existing EAI solutions. First, it provides enterprise

with a robust, flexible and sharable IT infrastructure.

Second, it achieves cross-platform, cross-boundary

and distributed sharing of the enterprise resources.

Third, it realizes the loosely coupled integration of

enterprise application.

In this paper, we carry out a simulation after we

introduce the EAI-Based-Grid solution and its

characteristics. The result of simulation proves the

solution of enterprise application integration based

on grid is superior to non-grid pattern.

However, the grid-related standards are still

being continually updated due to its immaturity.

Transforming the existing application system into

the grid service is still a hard work, which is also the

focus of our next step.

REFERENCES

Qureshi K A. 2005. Enterprises Application Integration.

Proceedings of the IEEE Symposium on Emerging

Technologies: 340-345.

Deng Wu, Xinhua Yang. 2008. Study on EAI Based on

Web Services and SOA. 2008 IEEE International

Symposium on Electronic Commerce and Security:

95-98.

Ning Zhang, Hong Bao. 2008. Study on Grid Technology

and its Application. 4th International IEEE

Conference on Wireless Communications, Networking

and Mobile Computing: 1-4.

Xiaoyong Gong, Qingsheng Zhu. 2006. Design and

Implementation of Enterprise Application Integration

Architecture Based on Grid Technology. Application

Research of Computers, Volume (12): 239 -241 (in

Chinese).

Yi Gu, Shensheng Zhang, Xiangfei Zhu. 2004. A Study of

Enterprise Appication Integration Solution. Computer

Engineering and Applications, Volume (6): 209-212

(in Chinese).

Iosup A, Dumitrescu C, Epema D. 2006. How are Real

Grids Used? The Analysis of Four Grid Traces and Its

Implications. 7th IEEE/ACM International Conference

on Digital Object Identifier: 262-269.

I Foster, C Kesselman, S Tuecke. 2001. The Anatomy of

the Grid: Enabling Scalable Virtual Organizations.

Journal of Super Computing Applications.

Benson E, Wasson G., Humphrey M. 2006. Evaluation of

UDDI as a provider of resource discovery services for

OGSA-based grids. 20th IEEE Interl. Symposium on

Parallel and Distributed Processing.

Bo Wang, Weiqin Tong, Wu Zhang. 2004. Service

Organization in OGSI. The 8th International IEEE

Conference on Computer Supported Cooperative

Work in Design: 469-474.

ICISO 2010 - International Conference on Informatics and Semiotics in Organisations

298