DOC PREVIEW
UT CS 395T - Writing Grid Service Using GT3 Core

This preview shows page 1-2-3 out of 8 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 8 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 8 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 8 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 8 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

Writing Grid Service Using GT3 Core Dec, 2003 Long Wang [email protected] Department of Electrical & Computer Engineering The University of Texas at Austin James C. Browne [email protected] Department of Computer Science The University of Texas at Austin Abstract OGSA (Open Grid Services Architecture) is proposed in 2002 which integrates the Globus standard with the commercial Web Service standard, provides a unified and standard architecture for grid computing. Grid service is the central concept of OGSA, which essentially is web service with improved functionalities. GT3 (Globus Toolkit 3.0) is the first implementation of OGSI (Open Grid Service Infrastructure) Specification 1.0. We show the usefulness of Grid Service by introducing a demo application based on GT3 core, and the step-by-step procedure of building Grid Service based application. Key Words: GT3, Grid Service1 1 Introduction Grid computing technology is being developed to solve two kinds of problems. First, there is much resource wasting in the internet. Such resources include processing cycles, disk space, data and network. Second, the integration of different systems deployed in a large company tends to be difficult. We need standard technology and platform to support such integration. To solve this problem, grid computing considers all the available resource in the network as a “super computer”. User can transparently use and manage all these resources. Grid computing also provides a series of standard to integrate heterogeneous systems. In 2002, OGSA (Open Grid Service Architecture) is proposed which integrates the Globus stand with the Web Service standard, and is going to be the unified standard for the grid computing. The basic concept of OGSA is grid service which is essentially a web service with improved functionalities and behaviors. Web service is selected because compared to other distributed computing technologies, such as CORBA, RMI and EJB, it is a more suitable candidate for internet scale application. First, web service is based on a collection of open standards, such as XML, SOAP, WSDL and UDDI. It is platform independent and programming language independent because it uses standard XML language. Second, web service uses HTTP as the communication protocol. That is a big advantage because most of the Internet’s proxies and firewalls will not mess with HTTP traffic. However, web service is not powerful to build complex applications. It lacks some functionality, such as lifecycle management, notification and persistency. And web service is stateless which means it can’t remember what has been done from one invocation to another. Grid service provides more versatile functionality than web service. We will cover these functionalities in section 2. In section 3, we will use a demo application to introduce how to develop grid service application by using GT3 core. In section 4, we will evaluate the grid service and GT3 core in terms of performance, functionality and ease of use. Section 5 will concludes the report and describes the future work. 2 Grid Service 2.1 Persistent Grid Service and Transient Service Instance2 Instead of having a single stateless web service instance shared by all the users. Grid service makes use of factory approach. For any grid service, a factory will be generated when this service is deployed in the server the first time. And this factory will in charge of all the service instances. For example, when a user needs to be served by a new service instance, it will talk to the factory and ask it to generate a new instance. Whenever a new instance is created, a global unique handle (GSH, Grid Service Handle) will be assigned to it. GSH is just like an address which will specify where the grid service is. To find out how to communicate with the grid service, the GSH is sent to the handle resolver, and the GSR (Grid Service Reference) will be returned which tells detail information about grid service, such as methods and message types. If SOAP binding is used, such GSR is just a WSDL file. Each service instance is transient, and its lifecycle can be managed by the users. We will talk about the lifecycle management later. In this way, different users can talk to individual service instance and will not mess with each other. Actually, one service instance can also serve multiple different users when needed. 2.2 Service Data We know in uddi directory, we can discover the service that can satisfy our requirements. But if there are many similar services, how can we know which service is most suitable. One possible solution is use WSDL since it is the language used to describe service. However, there are two problems using WSDL. First, it is too technical and not easy to query. Second, WSDL file is not easy to be modified dynamically if these service data need frequent change. Service Data is a structured collection of information that is associated to a grid service. Each service instance has its own service data set which includes one or multiple SDE (service data elements). A SDE can be a simple name-value pare, or a complex data structure. In the real implementation in GT3 core, each SDE is defined as a Java class, just like a Java Bean. We can use the similar methods to set and get the values of SDE. A SDD (service data description) file is used to define the data structure of the SDE which will be sent to client with the WSDL file so that client will know what kind of the service data this service has. 2.3 Lifecycle Management3 Since each service instance is transient, there needs to be some lifecycle management mechanisms. The lifecycle of the service instance refers to the time between instance creation and destruction. Some service instance may live as long as server, that means they will be loaded whenever the server restarts. There are two approaches provided by GT3 core to define the lifecycle of a service instance. One is to negotiate with service factory about the lifecycle before creating the instance. The other is to send some message to the service instance in the middle of its lifecycle. Such messages include requestTerminationBefore and requestTerminationAfter. GT3 core provides some callback methods to allow


View Full Document

UT CS 395T - Writing Grid Service Using GT3 Core

Documents in this Course
TERRA

TERRA

23 pages

OpenCL

OpenCL

15 pages

Byzantine

Byzantine

32 pages

Load more
Download Writing Grid Service Using GT3 Core
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Writing Grid Service Using GT3 Core and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Writing Grid Service Using GT3 Core 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?