GRIDSpace: Semantic Grid Services on the Web Evolution towards a SoftGrid Oct 29 – 31 2007 The International Conference on Semantics, Knowledge and Grid,

GRIDSpace: Semantic Grid Services on the Web — Evolution towards a SoftGrid

Oct 29 – 31 2007

The International Conference on Semantics, Knowledge and Grid, Xi’ an, China

Tharam S. Dillon, Chen Wu, Elizabeth ChangDEBI Institute, Curtin University of Technology, Australia

2

Agenda

Introduction

Overview of Technologies Grid Service Semantic Grid Web2.0

Motivation

GRIDSpace Conceptual Model Overall Architecture

Conclusion and future work

3

Introduction

Distributed computing Grid computing

－ Open Grid Services Architecture (OGSA) － Semantic Grid

Web applications SOA

－ IDC estimates that by 2010, 80% software profits from SOA

Observation Three areas of research are separate

－ Grid Services, Semantic Grid, and Web2.0

Goal Towards an organic Grid Ecosystems “on the Web”

4

Grid Services

OGSA Grid service: a potentially transient Web service based on

Grid protocol expressed by WSDL (Comito et al. 2005) “Transient” is the key

－ Grid: Virtual resource pool－ Web services: End-to-end Functional invocation

Grid service requires extra “virtual” management－ E.g. service instantiation, state management, etc.

provides Infrastructure Services, Data Services, Resource Management Services, Execution Management Services, Security Services, Self Management Services, and Information Services

Alternative to OGSA, previous studies proposed: Grid-based Web services Web-based Grid services

5

Semantic Grid

Semantic Grid an extension of the current Grid in which information

and services are given well-defined meaning (Roure et al. 2005)

But how it can be realised?－ Semantic Web + Semantic Web Services－ Semantic Grid reference architecture (Corcho et al.

2006)» Principles for migrating from OGSA to Semantic Grid

－ Web Services Modelling Ontology and Web Services Modelling Language (Toma et al. 2006)

－ Semantic Space (Zhuge and Li 2007)

6

First, Grid resources must be formalised into services based on open standards,

Second, service composition is the key for applications and users

In current Web2.0 settings, service composition comes as the Mashup, Web applications that combine information from several sources and is provided through simple Web APIs. In Grid computing settings, we believe Mashup plays a key role in forming the Virtual Organisation,

Third, Grid users at varying levels (i.e. end users, scientists, professionals) shall all contribute to the Grid platform during their participation.

7

Adopting both the service-oriented and user-oriented strategy to approach this level of service state management gives rise to the notion of “Transient Virtual Organisation”, which involves the lifecycle management of self-organised, dynamic Service Space

The existing Web services architecture based on Remote Procedure Call (RPC) is not indeed “Web-oriented”, raised concern that “Web services do not have much in common with the Web” [16], and has potential weakness such as scalability, performance, flexibility, and implementability [17].

GRIDSpace architecture is motivated by four major factors – Soft aspect of Grid, service-orientation, transient service space, and Web-oriented architecture – that are built on top of Grid services, Service-Oriented Computing, Semantic Grid, and Web2.0.

8

Web 2.0

Web 2.0 is an attitude or philosophy E.g. Using Web 2.0 for Service discovery (Wu and Chang 2007)

to think of Web as a platform of services Applications are built by composing services Users add “extra value” to these services through deeper

level of participation for their own benefits

Creating values from making connections Mashup is the key for value creation Mashup helps to form Virtual Organisation We differentiate

－ Service mashup (data level)－ User mashup (meta-data level)

9

It provides infrastructure to enable service discovery and “mashup” at various levels.

Grid services within a Service Space are referred to as ‘members’ of that Service Space.

In GRIDSpace, we have identified and instantiated three types of Grid Service Spaces: OGSA Space, Semantic Grid Space, and Virtual Organisation

10

Thinking of Web as services

“If you hit the Amazon.com gateway page, the application calls more than 100 services to collect data and construct the page for you” (Gray, 2006)

－ Werner Vogels, Amazon CTO

11

Thinking of Web as a Value Chain

Many vivid examples “on the Web” Google eBay Amazon

12

Motivation of GRIDSpace

Negligence of the “soft” aspect of the Grid computing, root cause: Resources sharing vs. Service orientation Resource: provided on the ‘as is’ basis, thus needs infrastructu

re to ‘push’ to the users Service: provided on the ‘demand’ basis and has been ‘pulle

d’ from low level components or resources OGSA attempts to bridge the gap through ‘transformation on th

e surface’ But we need radical solution to “soften” Grid

Incrementally value adding is the key We envision a Grid that supports hardware, software, and softwar

e/hardware hybrid resources It expands from the ‘middleware’ to a truly ‘SoftGrid’

13

Semantic Grid Space (SGS) refers to a focused Service Space where a group of related Grid services forms a domain-specific Grid service community in order to facilitate dependable collaboration through trust-driven service selection and semantic-based service discovery.

Transient Virtual Organisation (VO) is a demand-driven Service Space that allows a small group of Grid services to form an ad hoc team working collectively in order to fulfil particular user demands during a given period of time.

ad-hoc Grid service mashup – Grid service mediation, expansion, customisation, and integration are essential for a VO to satisfy real-world user requirements.

14

Conceptual Model for GRIDSpace

OGSA Space

Semantic Grid Space

Virtual Organisation

Grid Middleware Fabric

Applications

Grid Service

Grid Layer

Application Layer

Semantic Grid

Service Space Layer

Web

2.0

15

Architectural Design for GRIDSpace

16

Scientists can apply two approaches to semantically enrich existing Grid services.

The first top-down approach is based on the concept of ontology engineering [22], where scientists and domain experts manually annotate relevant Grid services using specific domain ontologies and/or knowledge databases.

The second empirical approach builds on practical methods such as data/text mining, business intelligence, machine learning that can be carried out (semi-) automatically without intensive human involvement.

17

Semantic Grid Space nurtures Grid services mainly through three means: semantic enrichment, semantic classification, and semantic discovery

rather than relying on the Service Broker, end users can also track down constantly-changing Grid services in any Service Spaces through the user-centred Grid Service Portal (GSP).

A GSP refers to a locally-accessible and highly-customisable user interface that provides a personalised view of activities and information essential to performing Service Space functions.

GSP acts as a proxy on behalf of the end users to maintain a list of communication channels to involved Service Spaces. Unlike a traditional HTTP proxy server shared by a group of corporate users, SSP is dedicated to serve only one user, thus creating the ‘user-centred’ view. SSP also reveals the notion of ‘User Mashup’– a core concept underpinning the attitude of Web2.0 [9].

18

Service Value ChainGlobal

SpaceDomain Syndication

Dynamic Alliance

19

User Mashup (see Figure 3) in the context of GSP refers to an activity in which the user can ‘hack’ standard Service Space communication protocols, and hence extensively customises user interface or features based on his own preferences

User Mashup provides a powerful yet simple mechanism by which infinite ‘virtual’ syndications of Service Spaces can be created for each SSP.

A Virtual Syndication of Service Spaces is a fresh, highly filtered, and combinatory view of several Services Spaces within a GSP. It is created, customised, and solely owned by each individual SSP user and does affect other users

20

User-centred Grid Service Portal

21

Prototype

Some preliminary implementation work…

22

Global Space

WSDL URLs Distribution (size = 22,863) Google Valid12%

Google Invalid11%

GoogleUnfetchable

46%

Public Valid3%

HTML28%

Sparse matrix A for Grid Services Collection

23

Preliminary Prototype

Grid Service Syndication Semantic Grid

Semantic

Space for

WSDL

A Cluster

Sim = 0.24

24

Preliminary Prototype

Grid Service User Mashup and Grid Portal

25

Future Work

User-Centred SoftGrid Infrastructure A Grid that can be used by users in daily life

Application User provides/consumes values:

－ Data－ Computation－ Intelligence

Infrastructure provides:－ Connection－ Search－ Mashup－ Management

26

Conclusion

Grid computing is about large-scale resource sharing OGSA strives to integrate Grid with SOA We advance the software middleware in OGSA towar

ds the notion of software/hardware/hybrid This leads to the GRIDSpace conceptual model and a

rchitectural design We envision a truly SoftGrid in the near future

27

About Tharam S. Dillon

PhD in Electronic Engineering from Monash University

Professor of Digital Ecosystems and Business Intelligence Institute at Curtin University of Technology

Was Dean of Information Technology, University of Technology Sydney

Was Chair Professor of the Department of Computer Science and Computer Engineering at La Trobe University

Was Head of the School of Engineering at La Trobe University

Was Professor of Computing and Acting Head of the Department of Computing at Hong Kong Polytechnic University

Research interests: Distributed Computing, Business Intelligence, Data

Mining, Neural Networks, XML-based Systems, Ontologies

Thank you!

http://debii.curtin.edu.au