WSRF-BASED VIRTUALIZATION FOR MANUFACTURING

RESOURCES

Lei Wu, Xiangxu Meng, Shijun Liu, Chenlei Yang, Xueqin Li

School of computer science and technology, Shandong University, Jinan, 250100, P.R.China

Keywords: Manufacturing Grid, manufacturing resources, resource virtualization, encapsulate template.

Abstract: Manufacturing Grid (MG) is a cooperation platform and resource sharing system among manufacturing

enterprises. To be shared in MG, the manufacturing resources should be virtualized and exposed as web

services. The paper presents a WSRF-based approach to virtualize resources conveniently in MG including

many resource encapsulation templates and a resource container. The design principle, the implementation

of the resource encapsulate template and the resource container are described in detail, where the resource

container is a WSRF service deployed in GT java core and can expose resources as WS-Resource. Through

these tools, resource providers can encapsulate their resources with the encapsulation template and deploy

them in the resource container, then the resource will be virtualized and exposed as web service and can be

shared in grid. A prototype platform to demonstrate the validity of the method is introduced at last.

1 INTRODUCTION

The principles of virtualization - the notion of

liberating functionality and management functions

from dedicated physical devices – have been around

for decades. Virtualization technologies encompass a

variety of mechanisms and techniques used to

decouple the architecture and user-perceived

behavior of hardware and software resources from

their physical implementation. (Renato Figueiredo,

2005)Resource virtualization provides an abstraction

that allows applications to access physical resources

indirectly, without being tied directly to specific

physical resources (such as servers or clusters). This

virtualization software provides a common interface

for applications. (N_emeth, 2003) points out that the

essence of a Grid is the virtualization of resource and

the virtualization of the concept of a user. Grids

apply the concept of resource virtualization by

aggregating numerous compute resources (from

mainframes to desktops) and abstracting them into a

single, unified resource that provides, in effect, a

single virtual computer that can be accessed as

needed or on demand.

The term of manufacturing grid (MG) is the

applying of grid technologies on manufacturing,

which is really a cooperation platform and resource

sharing system among manufacturing enterprises.(

Liu S.J,2005) Manufacturing resources are facilities

involved in the entire product lifecycle, including

material resource, information resource, technique

resource, human resource, financial resource and so

on. Some manufacturing resources can be shared

through network easily, such as various information,

data, and software, which can be used in cooperation

work supported by MG. Software and computing

service can be called via the network by means of

Web services. But there are still many manufacturing

resources which cannot be shared directly through

network, such as materials, equipments and so on.

(Liu S.J,2004 ).To be shared, these resources need to

be described, expressed, encapsulated and deployed

in MG. Via virtualization, all manufacturing

resources are exposed as Web services, which

eliminates the heterogeneousness of resources and

shortens the distance between them.

Web Services are the technology of choice for

Internet-based applications with loosely coupled

clients and servers. That makes them the natural

choice for building the next generation of grid-based

applications. However, Web Services do have

certain limitations. Plain Web services are usually

stateless (even though, in theory, there is nothing in

the Web Services Architecture that says they can't be

stateful). This means that the Web service can't

"remember" information, or keep state, from one

invocation to another. So plain Web Services(as

132

Wu L., Meng X., Liu S., Yang C. and Li X. (2006).

WSRF-BASED VIRTUALIZATION FOR MANUFACTURING RESOURCES.

In Proceedings of the Eighth International Conference on Enterprise Information Systems - SAIC, pages 132-137

DOI: 10.5220/0002452301320137

 SciTePress

currently specified by the W3C) wouldn't be very

helpful for building a grid application. However,

Grid applications do generally require statefulness.

So, we would ideally like our Web service to

somehow keep state information (Borja

Sotomayor,2004). The WSRF specifications were

introduced in January 2004, which define the

WS_Resource construct, a composition of a web

service and a stateful resource described by an XML

document (with known schema) that is associated

with the web services porttype and addressed by a

WS_Addressing Endpoint Reference(EPR) [ IBM,

BEA and Microsoft]. The four WSRF specifications

being standardized in OASIS[OASIS] define how to

represent ,access, manage, and group WS-Resources.

(Rajaram Shivram Kumar,2004) adopts Web

Services Resource Framework (WSRF) as a major

mechanism for virtualization. Using WSRF,

manufacturing resource will be exposed as a WS-

Resource. However, this method has certain

limitations. Every resource must be virtualized as a

separate web service, which makes the management

of these services very complex. Meanwhile, most of

resource providers can’t hold the WSRF

specifications, so it is difficult for them to virtualize

their resources and deploy them in grid. We present

the concept of resource encapsulation template and

resource container. The resource container is actually

a WSRF service, which can adopt resources

implemented by Java, C and .NET language and

virtualized them as WSRF services.

The rest of this paper is organized as follows.

We first introduce the design principle of our

approach in Section 2. In Section 3 the resource

encapsulation template is presented. The resource

container is described in detail in Section 4. The

prototype system is outlined in section 5. Finally the

Section 6 concludes the paper and outlines some

future work.

2 DESIGN PRINCIPLE OF

RESOURCE VIRTUALIZATION

To be shared in MG, manufacturing resources need

be virtualized as web services. There are many

requirements as follows:

(1)Resource virtualization should make various

manufacturing resources be shared conveniently in

MG.

(2)Resource virtualization should make resources

offer their functionality absolutely, meanwhile, hide

the implementation detail.

(3)Resource virtualization should make the using

and management of resources more convenient.

(4)Resource virtualization should make the

searching of resource more easy and exact.

There are various virtualization methods. One

way is encapsulating resources as web services.

Manufacturing resources are different in express

form, management mode, sharing mode, using

approach, trade mode, and so on. So it is difficult to

manage and use the resources encapsulated without

uniform specification. According to above

requirements several resource encapsulation

templates are presented, which are the specification

and certain interface followed when encapsulating

resources’ logical function. It supports java, c and

.net program language. Resource providers can

encapsulate their resources using these templates. To

expose these resources encapsulated with our

template as web service and shared in grid, we

design a resource container, which is actually a

WSRF service deployed in GT service container. It

General service client

General service invoking engine

Resource container A Resource container B

encapsulate template

Manufacturing

resource

encapsulate template

Manufacturing

resource

Figure 1: the structure of MG.

encapsulate template

Manufacturing

resource

encapsulate template

Manufacturing

resource

WSRF-BASED VIRTUALIZATION FOR MANUFACTURING RESOURCES

133

can adopt resources encapsulated with the template

presented above and automatically virtualized them

as WSRF services. So resource providers can

virtualize their resources as web services through

implementing some interface. They can use some

common program language such as Java, C and

.NET. The resource encapsulate template and the

resource container will be described in detail in

below sections. The structure of the grid is shown in

Figure 1. Manufacturing resources are encapsulated

with the resource encapsulation template and added

in one of resource containers. Resource container

can be deployed in any node of the grid. There are

two resource containers in our prototype. A general

service invoking engine is developed, which can

invoke resources deployed in different resource

containers to complete the users’ task.

3 RESOURCE ENCAPSULATION

TEMPLATE

Resource encapsulation template is some

encapsulation specifications and certain interface

that resource providers should implement. The

resource container supplies resources encapsulated

by diverse encapsulation templates based Java, C

and .NET. A template based java is presented in

detail in this paper. When resource providers

encapsulate their resources, they must implement the

interface as follows.

Public interface IMGResource{

Public void initResource(IResourceRefrence ref);

// this method is used to initialize the resource,

‘ref’ is the reference of the resource .the resource

state will be set in this method

Public HashMap getTaskInputParameters();

// return the input parameters of the task.

Public HashMap executeTask （ HashMap input

）;

// execute a task. All the logical function that the

resource offer is implemented in this method, the

result will be return in HashMap structure.

Public file getResDescription （）;

// get some description of the resource .we set

xml schema for every kind of resource ,the

description of the resource will be described with

xml document .

}

Resource providers implement the above

interface to encapsulate the resources’ logic

functions and deploy them in the resource container,

and then the resources can be exposed as WSRF

services and offer their functionality in a standard

Web service environment to be shared in grid.

4 RESOURCE CONTAINER

To expose resources encapsulated with our template

as web services, resource container is designed and

developed.

4.1 Design Principle of the Resource

Container

The resource container is actually a WSRF service. It

can be deployed in GT service container

conveniently. It can adopt resources encapsulated by

Resource

Manufacturing

resource

GT4

JAVA

CORE

WSRF

Service

Resource Containe

Equipment :machine tool

JavaClass：Device

Analyse software：ansys

JavaClass：Ansys

Design software：CAD

JavaClass：CAD4J

Manufacturing

resource

Manufacturing

resource

Figure 2: Architecture of Resource Container.

ICEIS 2006 - SOFTWARE AGENTS AND INTERNET COMPUTING

134

above template as it’s resources and expose them as

web services shared in grid.

The architecture of the resource container is

shown in Figure2.

The encapsulated resources are deployed in the

resource container, at the same time the information

of the resource including the address of the resource

container and the resourceID will be registered in a

private UDDI. The registry gets the description of

the resource, which is a facility for resource search.

A general resource invoking engine is designed

in the job manager, which can invoke resources

deployed in the resource container. When users want

to use some resources in grid, they firstly search the

resources by keywords from private UDDI registry.

Resources satisfying the requirement will be offered

by the service invoking engine. Then the invoking

engine will invoke the selected service to complete

the task.

4.2 How the Container Support

Resource Virtualizing

(1) Hot resource deploy

The encapsulated resources can be hot deployed in

the resource container. The resources can be used

without stopping the server container. If the

container has to restart to use the new resource, the

task running will be break in, which is not allowed.

So hot resource deploying is very significative.

(2) Remote resource deploy

The remote deploying interface is provided, the

resource providers can encapsulate their resources

with the resource encapsulation template, deploy

them in resource container and register into MG

from portal, which make the resource providers

virtualize and deploy their resources by themselves

conveniently

(3) Easily deployed and used

The resource container is actually WSRF service,

which can be easily deployed in some domain of the

grid with the “globus-deploy-gar” command. It

doesn’t need peculiar configuration. All resources in

one resource container can be deployed in another

resource container as a whole.

(4) Hiding implementing information

The encapsulated resources hide the

implementation information of the resources and

only the function interface is exposed to users, which

increases the security of resources.

(5) Self-describing of resource

Different xml schemas are defined for every kind

of resources. Resources are described with the xml

document. Resources’ description information can

be set when they are initialized.

(6) Effectively monitor

The resource container provides the

getResState() interface ,which ensures that the

application’s state and other data available .thus job

invoking engine can monitor running applications to

be able to fulfill an accepted SLA.

4.3 Implementation of the Resource

Container

There are many difficulties in designing the resource

container. This section describes the particular

design of the resource container in detail. Resource

container is a WSRF service, resources encapsulated

with above template will be added into the resource

container as WS-Resource. The particular design of

the resource container will be described in detail in

the next part of the section.

(1) Resource structure

MGRes{

String type;

String resName；

String className,

String resState；

IMGResource aResource；

}

The className is the name of the resource

implementation class, which must implement the

provided interface. ’resState’ is the state of the

resource, which is set by the resource implement

class. The resource invoking engine can monitor the

resource by getting this attribute. ’aResource’ is the

reference of the resource logical function. It points to

the resource implement class.

(2) Permanence of the resource

The information of the common WS-Resource

will be lost when the service container restart. To

store this information enduringly, the paper describes

the way on serializing resource. The resource

information will be saved in files. When the service

container start, the file will be read to unserialize the

resources, meanwhile the resource will be initialized.

The state of the resource will be reseted. The

resource file will be modified when resources are

added, removed and modified. Only the type,

resName and className will be serialized in file.

The ‘resState’ and ‘aResource’ need to be set when

the resource is initialized.

(3) Dynamic Load

WSRF-BASED VIRTUALIZATION FOR MANUFACTURING RESOURCES

135

When the resource is serialized, ’aResource’ is

not serialized. So the implement class must be

dynamicly loaded and set the ‘aResource’, make it

point to the implement class.

The dynamic load program as follows：

try {

IMgResourceImpl myObj=null;

Class myClass=null;

// get Jar file and classname

jarFileDir=C:/ws-core-4.0.0/bin/Temp;

jarFileName=” MgResourceImpl.jar”;

String jar=jarFileDir+"/"+jarFileName;

// Create URL for URLClassLoader

URL[] urls = {new URL("file:///" +

jar)};

// get the ClassLoader

ClassLoader

cl=Thread.currentThread().getContextCla

ssLoader();

URLClassLoader classLoader = new

URLClassLoader(urls,cl);

// Find the class

myClass =

classLoader.loadClass(className);

// Create an instance of the driver

myObj=

(IMgResourceImpl)myClass.newInstance();

resourcePackageResource.SetAResource(my

Obj);

//initialize the resource refrence.

ref=(IResourceRefrence)resourcePackageR

esource;

myObj.initResource(ref);

}catch(Exception e){

e.printStackTrace();

}

5 PROTOTYPE

We have developed the middleware to support the

establishing of a manufacturing grid system. A

prototype platform has been deployed to validate our

idea. The portal (www.mgrid.cn) can be visited now.

(Figure 3) Resource containers were deployed in two

nodes. The resource provider can download the

resource encapsulation templates from the portal,

encapsulate their resources and deploy them in the

resource container in the portal. Meanwhile the

virtualized resources are registered in a private

UDDI registry in the MGRID. The deploying and

resource can be shared in grid.

A kind of CAE software were virtualized with

above resource encapsulation template and deployed

in the resource container of our MGRID. Resource

users

can request this service through a invoking form

from portal. Figure 5 shows the form user should fill out.

6 CONCLUSIONS

Virtualization of manufacturing resources and

exposing them as web services is an ideal method of

manufacturing resources sharing. This paper presents

a virtualization method based on WSRF. The

resource encapsulation templates based java and the

resource container were introduced and developed.

Resource providers can virtualize their resources as

WSRF services conveniently and deploy them in

network. A prototype has been developed to validate

our idea. Various manufacturing resources have been

virtualized and shared conveniently in our MG

system now.

ACKNOWLEDGEMENTS

The authors would like to acknowledge the support

provided for the Science & Technology

Development Projects of Shandong Province

(031110147，2004GG1104011, 2004GG1104017)

ure 3: the

ortal of MGRID.

ICEIS 2006 - SOFTWARE AGENTS AND INTERNET COMPUTING

136

,Natural Science Foundation of Shandong

Province(Y2004G08), National Natural Science

Foundation of China(90612021)and the Research

Fund for the Dectoral Program of Higher

Education(20050422019).

REFERENCES

Renato Figueiredo, 2005. Resource Virtualization

Renaissance. In Computer Volume 38 , Issue 5 p28 -

31 .

N_emeth, Z., Sunderam, V,2003. Characterizing Grids:

Attributes, Definitions, and Formalisms. Journal of

Grid Computing

Liu S.J, Meng X.X, Tong Y.X，2005. SOMG: A Service

Oriented Manufacturing Grid System. In the Fifth

International Conference on Computer and

Information Technology.CIT'05 Press.

Liu, S.J, Meng, X.X, Gong, B, 2004. Resource Modeling

in Manufacturing Grid. In the 11th ISPE International

Conference on Concurrent Engineering: Research and

Applications.

Borja Sotomayor, 2004. The Globus Toolkit 4

Programmer's Tutorial University of Deusto.

IBM, BEA and Microsoft. 2004. WS_Addressig

http://msdn.microsoft.com/webservices/default.aspx?p

ull=/library/en-us/dnglobuspec/html/ws-

addressing.asp

David Snelling, 2004. OASIS Web Services Resource

Framework(WSRF) TC .http://www.oasis-

open.org/committees/tc_home.php?wg_abbrev=wsrf

Rajaram Shivram Kuma, 2005.Virtualization for

Manufacturing Web Services: a WS-RF approach .In

International journal of information technology

Volume 11, Number 3

Figure 4: resource deploying and register.

Figure 5: the form of the CAE service.

WSRF-BASED VIRTUALIZATION FOR MANUFACTURING RESOURCES

137