Transcript Middleware for Grid Computing and Relationship to

Middleware for
Grid Computing
and the
relationship to
Middleware at large
ECE 1770 : Middleware Systems
By: Sepehr (Sep) Seyedi
Date: Thurs. January 23, 2003
What is Grid Computing?
Definition of the Grid



Controlled sharing of geographically distributed
resources
Virtual Organization
Analogy with power grids

Consistent, pervasive, dependable, and transparent
access to electricity from different sources
Characteristics of a Grid




No centralized control center
Heterogeneity (of resources)
Scalability
Dynamic and Adaptable
Applications
Presented early to see relevance to the
design methodologies in grid middleware
Classifications of Applications/Uses:

Computational Service


Data Service


Example: like web services
Information Service


Scalable storage and access to distributed datasets
Application Service


Inherent part of ALL applications
Example: WWW portal
Knowledge Service

Example: data mining
A Layered View of a
Grid Grid Application
Available as APIs:
global resource
management: resource
brokers, …
User-level Middleware
Core Middleware
Grid Fabric
Process management,
co-allocation of
resources, storage
access, security,
information registry and
discovery, ….
Computers,
supercomputers,
storage devices,
instruments …
Middleware Overview
Many organizations involved

ex: Global Grid Forum and Globus
Components

Security


Secure communications (SSL)
Distributed security infrastructure



Transferring data throughout the grid and to users
Globus: GridFTP


Deals with high-performance, security and reliability
Information Management



Manage user credentials to selectingappropriate resources
Data Management

User-level
&
Core
ASPECTS
Information DB about resources
Availability, capabilities, …
Resource Management
Resource Management
Application
Resource
Resource
Broker
Broker
job
Multi-request
Query
? Types of
resources needed ?
 Specifications of
requests
Resource
Information Manager
Resource Co-allocator
request
request
request
Local Resource
Manager
Local Resource
Manager
Local Resource
Manager
Service/Resource
Service/Resource
Service/Resource
Open Grid Services
Architecture (OGSA)
Global Grid Forum (www.gridforum.org)
driving force behind OGSA
Promoting a standard and open
architecture for interfaces to resources
Assuring proper uniform behavior of all
resources and interoperability in a
heterogeneous environment
The Grid Service
Implemented as special a Web service
Standardized behavior  Standardized
interfaces
Standard interfaces in terms of Web Services
Description Language (WSDL)  GSDL
Implementation independent


The grid service is able to live on any host
environment that supports web services.
Examples: J2EE, .NET, C, Python, … which cover
most existing operating systems
The Grid Service Interface
Grid Services must support:

Discovery


Dynamic Service (instance) creation



Factory: creates service instances
Lifetime management


Registry: holds information about service instances
Life length, termination of service in case of failure of
resource
Notification
Manageability

Monitoring and managing service instances
Grid Services can also provide (custom)
application services, like web services
How does a Grid Service
work?
Client uses a Grid service interface
A grid service instance is created from a
Factory with the help of a Registry
The grid service instances run with
appropriate resources automatically allocated
New instances can allocated and destroyed
dynamically, to benefit performance

Example: A web serving environment could
dynamically allocate extra instances to provide
consistent user response time
Invocation
Factory
Service Instance
Service Instance
Service Instance
Real
Services
DB 1
…
Simple Invocation
Example
User
Database
Service
Database
Service
DB n
Service Instance
Service Instance
-EJB
-Java Class
Service Instance
Service
Instance
Service
Instance
Service Instance
Registry
Compute service provider
…
Hierarchical Use of
Grid Services
Simple Hosting
Environment
Virtual Hosting
Environment
Collective
Services
Environment
Relationship with
Middleware at large
Distributed Computing



Grid computing itself can be thought of as a
distributed system
Differs by providing a controlled, dynamic
resource-sharing framework that addresses the
requirements of a grid
Comparison to CORBA, J2EE, and .NET
Grid computing can be used as a building
block in other middleware systems

‘Too much middleware’
Benefiting other middleware technologies

Web services is making advances due to grid
services
Conclusions
Characteristics of a grid relevant to
middleware
Common design methodologies in grid
middleware
Grid Services and open standardization
New and existing middleware systems are
beginning to adopt core grid middleware to
become easily ‘grid-enhanced’
Thank You!
Any Questions ??
CHECKOUT
www.gridcomputing.com
for more information on
Grid Computing!