Concurrency control and recovery in database systems
Concurrency control and recovery in database systems
Distributed programming in Argus
Communications of the ACM
Dynamic voting algorithms for maintaining the consistency of a replicated database
ACM Transactions on Database Systems (TODS)
Experience with transactions in QuickSilver
SOSP '91 Proceedings of the thirteenth ACM symposium on Operating systems principles
SIGMOD '92 Proceedings of the 1992 ACM SIGMOD international conference on Management of data
An efficient, fault-tolerant protocol for replicated data management
PODS '85 Proceedings of the fourth ACM SIGACT-SIGMOD symposium on Principles of database systems
Fail-stop processors: an approach to designing fault-tolerant computing systems
ACM Transactions on Computer Systems (TOCS)
SIGMOD '81 Proceedings of the 1981 ACM SIGMOD international conference on Management of data
A Quorum-Based Commit and Termination Protocol for Distributed Database Systems
Proceedings of the Fourth International Conference on Data Engineering
The Grid Protocol: A High Performance Scheme for Maintaining Replicated Data
Proceedings of the Sixth International Conference on Data Engineering
Weighted voting for replicated data
SOSP '79 Proceedings of the seventh ACM symposium on Operating systems principles
A Transaction Replication Scheme for a Replicated Database with Node Autonomy
VLDB '94 Proceedings of the 20th International Conference on Very Large Data Bases
The circular two-phase commit protocol
DASFAA'07 Proceedings of the 12th international conference on Database systems for advanced applications
Hi-index | 0.00 |
When failures occur during the execution of distributed commit protocols, the protocols may block in some partitions to avoid inconsistent termination of the transaction, thus making data items in these partitions unavailable for accesses. We present a protocol that incorporates two new ideas with the goal of improving data availability. First, a new two-level voting scheme is proposed for deciding in which partitions to terminate the transaction. In this scheme, a choice is made based on the number of data items available in the partition rather than on the number of individual nodes. Indeed, in replicated systems, a criterion based on the number of nodes may be misleading. Second, we propose a way to reduce blocking caused by accumulating network fragmentation. The idea employs the views mechanism previously used in replica management.