GRID COMPUTING: REPLICATION CONCEPTS

1
GRID COMPUTING REPLICATION CONCEPTS

• Presented By
• Payal Patel

2
Outline

• What is Grid computing?
• Replication in grid computing
• Creation of replica in Grid computing
• File level replication
• Object level replication
• Selection of replica
• Deletion of replica
• Conclusion

3
What is Grid Computing?

• Aim share and distribute data, storage and
  other resources.1
• It provides
• global sharing of distributed resources
• non-trivial qualities of service.2
• businesses can optimize computing and data
  resources.2
• reliable and transparent access

4
Grid computing (cont.)

• Component of Grid computing.2
• Security
• Data Management
• Resource Management
• Information Services

5
Grid Computing (cont.)

• Figure from 1

6

Replication in grid computing

• Replication means copies of data in different
  locations.
• Goal To provide user to access data anywhere if
  any failure occurs.
• Improves data access performance.
• Grid computing uses Dynamic replication

7
Creation of replica

• Goal To achieve
• high data availability
• Fault tolerance
• Creation of replica depends on granularities of
  levels.5
• File Level
• Object level

8
File Level Replica Creation

• Grid Data Management pilot (GDMP)
• Used as file replication software system
• It has three components 1
• Replica catalog service
• Three levels of catalog
• Collection
• location
• Logical file entry
• Data mover services
• The storage manager service

9
Conti.

• Figure from 1

10
Object Level Replication

• Used when file level replication uses large
  amount of bandwidth and disk space.
• Figure from 1

11
Selection Of Replica

• Selection criteria1
• Discovering physical location of a file given a
  logical file name.
• Replica Location Service (RLS)
• Selecting the best replica from a set of physical
  replica location.

12
Deletion of Replica

• Maintain efficient resource utilization
• Free up unused storage.
• Uses decentralized replica deletion algorithm
• Replica deletion agent (RDA) to delete replica.1

13
Conclusion

• Grid computing provides large scale resource
  sharing and high performance orientation.
• File level and object level are two types of
  replica creation
• I also discussed selection and deletion criteria
  of replica
• Latest ongoing research in grid computing
• Grid The next generation internet

14
References

• Fay Hui, Cheryl Kang and Ichitaro Yamazaki.
  Distributed Data Replication in Grid Computing.
• IBM grid computing website at http//www-1.ibm.com
  /grid/
• Globus project. http//www.globus.org
• Grid 2004 from rocket science to business
  service

15

• Thank you