Distributed Heterogeneous Data Warehouse For Grid Analysis

1

• Distributed Heterogeneous Data Warehouse For
  Grid Analysis
• Harvey B. Newman, Julian Bunn, Saima Iqbal
• CALTECH ( California Institute of Technology ).

2
OUTLINE

• Introduction
• What is Relational data warehouse
• Distributed Heterogeneous Relational Data
  Warehouse Databases (DHRD) and Grid
• How DHRD could be integrated with the Grid
• Why web services?
• Building blocks of Web Services
• Vital parts of Web Services
• How DHRD could be integrated with the Grid as a
  Web Service
• Grid services
• Grid services architecture (GGF) draft 16th Feb.
  2003
• Grid services client infrastructure (GGF) draft
  5th Jun. 2003
• Proposed web services architecture based on Grid
  services to use DHRD in Grid environment
• Technologies employed
• UDDI complaint registry service

3
INTRODUCTION

• Can databases integrated with the Grid ?
• Most of the existing and proposed Grid
  applications are file based.
• Very little work has been done on how Distributed
  Heterogeneous Databases can be made available on
  the Grid.
• Web Services can help in accessing Distributed
  Heterogeneous Databases as a single Virtual
  Database across the Grid.

4
Distributed Data Warehouse

• The distributed database system allows
  applications to access data from local remote
  databases.
• It helps to move some of data and some of the
  users to separate servers and databases.
• Allow to keep data by a particular workgroup at
  Tier 2 and Tier 3, on a server nearby.
• Reduce the need for massive central computing and
  network delays.

5
Distributed Heterogeneous Relational Data
Warehouse (DHRD) Databases and Grid

• Is it possible to access DHRD databases across
  Grid by adopting the existing Grid services that
  handle files?
• While relational databases offers much richer set
  of operations like queries and transactions.
• There is much differences exists among different
  DBMS as that of different file systems.
• Even within one paradigm different databases
  products ( Oracle, MS-SQL, DB2) vary in their
  functionality and interfaces.

6
How DHRD Could Be Integrated with The Grid

• The diversity of DHRD makes it difficult to
  design a single solution to integrate DHRD
  databases with Grid.
• The Open Grid Services Architecture (OGSA) for
  distributed system provide the concept of Grid
  Services (like Web Services) to access resources
  across distributed and heterogeneous environment.
• These Grid Services/Web Services can help in
  providing the distributed databases across the
  Grid as a Virtual Database System.

7
Why Web Services?

• Web Services are centered on the Service
  definition and messages
• Web Services build on set of well established
  technologies and protocols
• - XML used for service description and data
  interchange
• http used as a transport protocol
• - widely deployed with trusted security
  features
• Web Services standards are structured and
  extensible
• - Interface evolution without breaking what is
  already working
• Provide solution for the access of heterogeneous,
  web-wide resources.

8
Building blocks Of Web Services

• Web Services are modular software components
  wrapped inside a specific set of Internet
  communication protocols and that can be run over
  the Internet.
• At the heart, web services architecture is the
  need for program-to-program communications.
• Key roles in the web services architecture are
• - a service provider
• - a service registry
• - a service requestor

9
Building blocks Of Web Services (contd)

• - Together they perform three operation on web
  services
• Publish, Find and Bind

SERVICE PROVIDER
1 Publish
3 Bind
Make the service description publicly available
Allows the service to be used by the requestor
SERVICE REQUESTOR
SERVICE REGISTRY
2 Find
Discover the service
10
Vital Parts of Web Services

• SOAP (Simple Object Access Protocol) through
  which the service provider, service registry and
  service requestor communicate.
• WSDL( Web Services Description Language) is the
  language used to create service description.
• UDDI (Universal Description Discovery and
  Integration) is the directory technology used by
  service registries that contain the description
  of web services and allows the directory to be
  searched for a particular web service.

11
How DHRD Could Be Integrated with The Grid As A
Web Service

• The Distributed Heterogeneous Relational
  Databases can register themselves as a web
  service in a UDDI registry.
• These web services can be accessible by a client
  through web application by using WSDL.
• In this architecture Client is very important
  because this Client can dynamically discover
  services, configure the remote calls on the basis
  of the inputs it gets from http call.

12
Grid Services

• The OGSA integrates key Grid technologies
  (including Globus toolkit) with Web Services
  mechanisms to create a distributed system
  framework around the OGSI (Open Grid Services
  Infrastructure).
• A Grid Service is a Web Service that conforms to
  a set of conventions (interfaces behavior) that
  define how a client interacts with a services
  available across Grid.

13
Grid Services Architecture (contd) (Grid
Database Service specification (GGF))
GridDataService
GridServicePort
FindServiceData
ltServiceDatagt
GridDataServicePort
Perform
Requester
ltResponsegt
GridDataTransport Port
Put/get
ltResponsegt
GridDataService
Requester Using Grid Data Service Ports
14
Grid Services Architecture (Grid Database
Service specification (GGF))
GridServiceRegistry
FindServiceData
GSH(GridServiceHandler)
CreateService
Requester
GridDataServiceFactory
ltServiceInformationgt
create
GridDataService
Database Servers
Creating a Grid Data Service
15
Grid Services Client Infrastructure (Grid
Database Service specification (GGF))
Proxy
Binding Selection
Client Application
Protocol 1 (binding) Specific stub
Invocation of Web Service
A Client-Server Interface
Protocol 2 (binding) Specific stub
A Client-Side runtime architecture
16
Proposed Web Services Architecture Based on Grid
ServicesTo Use DHRD In Grid Environment
ORACLE9i SERVER DATA (META DATA)
MonaLisa
Data Replication through
SSL
(Service Registry)
UUDI Registry Server
ORACLE9i SERVER DATA (META DATA)
 
JAVA XML API to connect with Database Server
Web Server
SOAP
HTTP Server
Server with Master Database
SOAP Processor
DISTRIBUTED DATABASE
WSDL file
(Service Provider)
UDDI SOAP Request and Response
SOAP
Bind with the provided service
MS-SQL DATA (META DATA)
Server with Materialized View Database
Client Web Application to connect with database
(Service Requestor)
17
Technologies Employed

• Java Web Services Developer Pack 1.0 (JWSDP)
• Apache Tomcat 4.1.2 for Java Web Services
  Developer Pack 1.0
• -Apache web server
• -Tomcat servlet engine
• Java API for XML Registries (JAXR) 1.0_02
• Java API for XML-based RPC (JAX-RPC) 1.0_01
• Web Application Deployment Tool for JWSDP
• XRPCC tool to generate WSDL
• JWSDP Registry Server 1.0_02
• -Xindice database, the repository for registry
  data
• -implements Version 2.00 of the Universal
  Description, Discovery and Integration (UUDI)

18
UDDI Complaint Service Registry

• A standardized, transparent mechanism for
  describing the service
• A simple mechanism for invoking the service
• An accessible central registry services
• Make use of XML and SOAP
• Provide service discovery platform on WWW
• Suitable for Black Box web environment
• Allow to store as much as detail about a service
  and its implementation as desired
• The UDDI version 2.0 API defines approx. 40
  messages to perform inquiry and publishing
  functions against any UDDI complaint service
  registry
• The schema defines 25 requests and 15 responses

19
Working of Web Services Prototype
6
SOAP Message
Program Implementation
Database Server
Ties
7
JAX-RPC Runtime
SOAP Message
8
5
JAX-RPC Runtime
Registry Server
JAXR
Web server
Stubs
JAX-RPC
3
Program Interface
2
9
4
Find-service
1
10
SOAP Message
http
Web Service Requester
20
Working of Web Services Prototype
21
Working of Web Services Prototype

• DEMO

22
Conclusion

• It seems possible that we can make the
  Distributed Heterogeneous Relational Data
  Warehouse Databases available across the Grid in
  form of Web Services/Grid Services.

23
Future Work

• Integration of MonALISA (Grid monitoring tool),
  for the location of required web service with
  optimal network resources
• Exploit UDDI with its full functionality
• Provide an API to integrate this Grid Services
  based Web Services prototype into the Globus
  toolkit

24

• Questions?