Horus: a flexible group communication system
Communications of the ACM
The dangers of replication and a solution
SIGMOD '96 Proceedings of the 1996 ACM SIGMOD international conference on Management of data
The implementation of a CORBA object group service
Theory and Practice of Object Systems - Special issue high availability in CORBA
A new approach to developing and implementing eager database replication protocols
ACM Transactions on Database Systems (TODS)
High Performance Cluster Computing: Programming and Applications
High Performance Cluster Computing: Programming and Applications
Don't Be Lazy, Be Consistent: Postgres-R, A New Way to Implement Database Replication
VLDB '00 Proceedings of the 26th International Conference on Very Large Data Bases
Improving the Scalability of Fault-Tolerant Database Clusters
ICDCS '02 Proceedings of the 22 nd International Conference on Distributed Computing Systems (ICDCS'02)
Partial Database Replication using Epidemic Communication
ICDCS '02 Proceedings of the 22 nd International Conference on Distributed Computing Systems (ICDCS'02)
Transactions on Partially Replicated Data based on Reliable and Atomic Multicasts
ICDCS '01 Proceedings of the The 21st International Conference on Distributed Computing Systems
Hi-index | 0.00 |
This paper proposes a dynamic partial replication protocol based upon group communication system for use with a web-enable database cluster. It dynamically combines the advantages of both a partial and a full replication model according to a query pattern. Most eager-update replication protocols that have been suggested as the best replication for a database cluster are based on the full replication. However, an actual database cluster system needs partial replication rather than full replication to achieve high throughputs and scalability. The proposed Dynamic partial Replication (DPR) protocol guarantees consistency among replicas and reduces the overhead due to remote access inherent in the previous partial replication protocols. The proposed protocol consists of three parts: partial replica control, scale-out factor estimation and dynamic replica allocation. Partial replica control part is the framework for the DPR protocol. Scale-out factor estimation part determines the optimal number of replicas according to the current query pattern and access frequency to maximize throughput and efficiency. Dynamic replica allocation part creates or removes the temporary replica in a local site. The simulated evaluation shows that the proposed protocol outperforms the existing eager-update protocols, achieving improvements of approximately 16% in response time and 20% in scalability.