1. Grid Computing
Sudhindra Rao

2. Outline History of Distributed Computing
Grid – Definition, Architecture details
P2P versus Grid
Webservices
Java – anywhere computing paradigm
Middleware
Grid models and recent research
Research directions
Tools and grids available
References

3. History
Shift from Centralized Computing to Distributed Computing – powerful processors, faster networks
Parallel computing based on MPI and PVM models
Cluster Computing
Peer-to-peer computing
Grid computing

4. Application and Infrastructure technology trends
Serial applications
Parallel applications
Multi-threaded
MPI/PVM
OpenMP
Client Server
CORBA
COM/DCOM
. NET
J2EE
Custom distributed systems
P2P
App Integration
Reliable messaging
Reliable execution
Service virtualization
Web services
Service Registration
Service discovery
Location independent service invocation
Lifting apps off the servers
Time
Monolithic
Open
Distributed
Virtualized
Mainframes
Storage :
Direct attached Storage
Open Systems
Unix
Windows
Linux
Clusters
DRM
Infrastructure Virtualization
Grid
OGSA
Data Grid
Service provisioning

5. Technology Evolution: Cluster, Grid, P2P
* Sputnik
1960
1970
1975
1980
1985
1990
1995
2000
* ARPANET *
* Ethernet
* TCP/IP
* IETF
* Internet Era
* WWW Era
* Mosaic
* XML *
PC Clusters
Crays
MPPs
Mainframes
* HTML
* W3C
P2P
Grids
*XEROX PARC worm
COMPUTING
NETWORKING
Minicomputers
PCs
WS Clusters
PDAs
Workstations
HTC
* Web Services

6. What is Cluster/Grid ?
A type of parallel and distributed system that enables the sharing, selection, & aggregation of resources distributed in administrative domains depending on their availability, capability, performance, cost, and users quality of service requirements.
Grid
A Cluster
A Single
Cluster

7. Approaches for Parallel Programming
Implicit Parallelism
Supported by parallel languages and parallelizing compilers that take care of identifying parallelism, the scheduling of calculations and the placement of data.
Explicit Parallelism
In this approach, the programmer is responsible for most of the parallelization effort such as task decomposition, mapping task to processors, the communication structure.
This approach is based on the assumption that the user is often the best judge of how parallelism can be exploited for a particular application.

8. Parallel Programming Models and Tools
Shared Memory Model
DSM
Threads/OpenMP (enabled for clusters)
Java threads (HKU JESSICA, IBM cJVM)
Message Passing Model
PVM
MPI
Hybrid Model
Mixing shared and distributed memory model
Using OpenMP and MPI together
Object and Service Oriented Models
Wide area distributed computing technologies
OO: CORBA, DCOM, etc.
Services: Web Services-based service composition 14

9. Levels of Parallelism PVM/MPI Threads Compilers CPU Task i-l Task i
Code-Granularity
Code Item
Large grain
(task level)
Program
Medium grain
(control level)
Function (thread)
Fine grain
(data level)
Loop (Compiler)
Very fine grain
(multiple issue)
With hardware
PVM/MPI
Task i-l
Task i
Task i+1
func1 ( ) {
.... }
func2 ( ) {
.... }
func3 ( ) {
.... }
Threads
Compilers
a ( 0 ) =..
b ( 0 ) =..
a ( 1 )=..
b ( 1 )=..
a ( 2 )=..
b ( 2 )=..
CPU + x
Load

10. Cluster Architecture Parallel Applications Parallel Applications
Sequential Applications
Sequential Applications
Sequential Applications
Parallel Programming Environment
Cluster Middleware
(Single System Image and Availability Infrastructure)
PC/Workstation
Network Interface Hardware
Communications
Software
PC/Workstation
Network Interface Hardware
Communications
Software
PC/Workstation
Network Interface Hardware
Communications
Software
PC/Workstation
Network Interface Hardware
Communications
Software
Cluster Interconnection Network/Switch

11. A Typical P2P Computing Environment
Peer Discovery Service
Peer Agent
Application P3 pM
Who can help ?
Peer P2, P7 can help! pN
Request P2
Sorry, I am busy.
Peer Agent
Request
Peer Agent
Response P1 R7 p4 p5

12. CPM: DC Economy-based P2P Computing (Jxta based Implementation)
Market Server
Market Repository
Discovery
- Membership
CPM Agent
User (Consumer)
Bill
Trader
Job Management
Resources
(Provider)
Accounting

13. Definition of a Grid
Grid is a type of parallel and distributed system that enables the sharing, selection, and aggregation of geographically distributed "autonomous" resources dynamically at runtime depending on their availability, capability, performance, cost, and users' quality-of-service requirements
Coordinated resource sharing and problem solving in dynamic, multi-institutional Virtual Organizations (VOs)
Most current distributed technologies facilitate this in a local environment
J2EE, CORBA, VPN are a few examples
Nomadic users and applications provide new avenues for providing such a service
Mechanisms required to coordinate trusted and untrusted access to resources

14. Grid Architecture

15. A Typical Grid Computing Environment
Grid Information Service
Grid Resource Broker
database
Application R2 2 R3 R4 R5 RN
Grid Resource Broker R6 R1
Resource Broker
Grid Information Service

16. Virtual Drug Design A Virtual Lab for “Molecular Modeling for Drug Design” on P2P Grid
Data Replica Catalogue
Grid Market
Directory
Grid Info.
Service
“Give me list PDBs sources
Of type aldrich_300?”
“service cost?”
“service providers?”
GTS
Resource
Broker
“Screen 2K molecules in 30min. for $10”
“mol.5 please?”
GTS
(RB maps suitable Grid nodes and Protein DataBank)
“get mol.10 from pdb1 & screen it.”
PDB2
GTS
“mol.10 please?”
GTS
GTS
(GTS - Grid
Trade Server)
PDB1

17. Scalable Seamless Computing: Breaking Administrative Barriers
2100 ?
PERFORMANCE
2100
Administrative Barriers
Individual
Group
Department
Campus
State
National
Globe
Inter Planet
Galaxy
Desktop
SMPs or SuperComputers
Local
Cluster
Enterprise
Cluster/Grid
Global
Cluster/Grid
Inter Planetary
Grid!

18. Basic Elements Application Development Tools Security
Uniform Access
Security
System Management
Computational Economy
Resource Discovery
Resource Allocation
& Scheduling
Data locality
Network Management
Application Development Tools

19. Cluster, Grid, P2P: Characteristics
Population
Commodity Computers
High-end computers
Edge of network (desktop PC)
Ownership
Single
Multiple
Discovery
Membership Services
Centralised Index & Decentralised Info
Decentralized
User Management
Centralised
Decentralised
Resource mgmt
Centralized
Distributed
Allocation/Scheduling
Inter-Operability
VIA based?
No standards yet
No standards
Single System Image
Yes No
Scalability
100s
1000?
Millions?
Capacity
Guaranteed
Varies, but high
Varies
Throughput
Medium
High
Very High
Speed(Lat. Bandwidth)
Low, high
High, Low

20. Issues in Grid computing
Protocols required for interoperability
Define standard services – for access of computation, data, resource discovery etc.
APIs and SDKs to assist such protocol and service deployment
Current Distributed Computing – Resource sharing in single organization – limited to sharing certain resource types only
Need of services to support a common set of applications – Middleware

21. Projects
Globus – A toolkit for grid computing infrastructure development
Gridbus
Legion
OGSA – Standard for developing Grid application infrastructure (derived from Globus)

22. Grid Computing Approaches
mix-and-match
Object-oriented
Internet/partial-P2P
Grid Computing
Approaches
Network enabled Solvers
NetSolve
Economic-based Utility / Service-Oriented Computing
Nimrod-G

23. Some Global Initiatives
USA
AppLeS
Globus
Legion
Sun Grid Engine
NASA IPG
Condor-G
Jxta
NetSolve
AccessGrid
and many more...
Cycle Stealing & .com Initiatives
Distributed.net ….
Entropia, UD, SCS,….
Public Forums
Global Grid Forum
Australian Grid Forum
IEEE TFCC
CCGrid conference
P2P conference
Australia
Nimrod-G
Gridbus
GridSim
Virtual Lab
DISCWorld
GrangeNet.
..etc
Europe
UK eScience
EU Data Grid
Cactus
XtremeWeb
..etc.
India
I-Grid
Japan
Ninf
DataFarm
Korea...
N*Grid
Singapore
NGP

24. Globus Approach
A toolkit and collection of services addressing key technical problems
Modular “bag of services” model
Not a vertically integrated solution
General infrastructure tools (aka middleware) that can be applied to many application domains
Inter-domain issues, rather than clustering
Integration of intra-domain solutions
Distinguish between local and global services

25. Grid computing – SuperScalar model IBM
Ease the programming of GRID applications
Basic idea:
Grid 
ns  seconds/minutes/hours

26. Automatic code generation
app.idl
gsstubgen
client
server
app.c
app-stubs.c
app.h
app-worker.c
app-functions.c

27. Automatic code generation
serveri
app-functions.c
app-worker.c
app.c
app-stubs.c .
GRID superscalar
runtime
GT2
serveri
app-functions.c
app-worker.c
client

28. Production Grids & Testbeds
NASA’s Information Power Grid
The Alliance National Technology Grid
GUSTO Testbed

29. Testbed Statistics (Browse the Testbed)
Grid Nodes: 218 distributed across 62 sites in 21 countries.
Laptops, desktop PCs, WS, SMPs, Clusters, supercomputers
Total CPUs: (~3 TeraFlops)
CPU Architecture:
Intel x86, IA64, AMD, PowerPC, Alpha, MIPS
Operating Systems:
Windows or Unix-variants – Linux, Solaris, AIX, OSF, Irix, HP-UX
Intranode Network:
Ethernet, Fast Ethernet, Gigabit, Myrinet, QsNet, PARAMNet
Internet/Wide Area Networks
GrangeNet, AARNet, ERNet, APAN, TransPAC, & so on.

30. Grid Technologies and Applications
Natural Language Engineering
High Energy Physics
Brain Activity Analysis
Grid
Apps.
Molecular Docking
Portfolio Analysis
GAMESS Chemistry
High-level Services and Tools …
User-Level Middleware
(Grid Tools)
G-Monitor
Programming Framework
Gridscape
Grid Brokers & Schedulers
Nimrod-G
Gridbus Data Broker
Alchemi:
.NET Grid Services
+Clustering of desktop PCs
Globus
Data Management Services
Grid Bank
GMD
Core Grid
Middleware
MDS
GRAM
GASS
PKI-based Grid Security Interface (GSI)
.NET
JVM
Condor
PBS
SGE
LSF
Tomcat
Grid
Fabric
Windows
Solaris
Linux
AIX
IRIX
OSF1
HP UX

31. Classes of Applications that can be powered by Grids
Distributed HPC (Supercomputing):
Computational science.
High-Capacity/Throughput Computing:
Large scale simulation/chip design & parameter studies.
Content Sharing (free or paid)
Sharing digital contents among peers (e.g., Napster)
Remote software access/renting services:
Application service provides (ASPs) & Web services.
Data-intensive computing:
Drug Design, Particle Physics, Stock Prediction...
On-demand, realtime computing:
Medical instrumentation & Mission Critical.
Collaborative Computing:
Collaborative design, Data exploration, education.
Service Oriented Computing (SOC):
Towards economic-based Utility Computing: New paradigm, new applications, new industries, and new business.

32. Analysis Summary Application Data Size Processing Time Nodes
Belle Analysis (HEP)
300 MB input (100 jobs – 3MB each)
30 min.
Australia, Japan
Financial Portfolio Analysis
50 MB output (50 jobs – 1MB each)
20 min.
Global
Newswire Indexing
80 MB input (12 jobs – 7MB each job)
GrangeNet, Australia
GAMESS
4KB for each job.
Total output: 860MB compressed
Each job took 5-78 minutes.
Total 15 hours
(130 nodes, 15 sites)

33. What is Grid computing? Grid is the next-generation internet
Grid requires a distributed operating system
Grid requires new programming models
Grid does not need high performance computers

34. Research directions
Publisher/Subscriber systems on the Grid – How can the grid be used to manage such applications and what are the issues
What levels of selectivity and regionalism is expected from VOs?
How to handle the dynamics of the topology and nodes?
Addressing QoS on Grid – best effort ?
Efficient Discovery and Retrieval
Replication techniques

35. References
List of available resources on grid computing -
Foster I., Kesselman, C., and Tuecke, S., - “The Anatomy of the Grid- Enabling Scalable Virtual Organizations” – Intl J. SuperComputer Applications, 2001
Casanova, H., “Distributed Computing Research Issues in Grid Computing” – ACM SIGACT News Distributed Computing Column 8 July, 2002
Lau, F., Ho, R. and Wang, C., “Grid Computing: Challenges and Design Approaches”
“The grid : blueprint for a new computing infrastructure” Editors Foster, I., and Kesselman, C. , Elsevier, 2004