Lazy replication: exploiting the semantics of distributed services
PODC '90 Proceedings of the ninth annual ACM symposium on Principles of distributed computing
Implementing fault-tolerant services using the state machine approach: a tutorial
ACM Computing Surveys (CSUR)
Understanding fault-tolerant distributed systems
Communications of the ACM
Providing high availability using lazy replication
ACM Transactions on Computer Systems (TOCS)
Replication and consistency: being lazy helps sometimes
PODS '97 Proceedings of the sixteenth ACM SIGACT-SIGMOD-SIGART symposium on Principles of database systems
Replication, consistency, and practicality: are these mutually exclusive?
SIGMOD '98 Proceedings of the 1998 ACM SIGMOD international conference on Management of data
A new approach to developing and implementing eager database replication protocols
ACM Transactions on Database Systems (TODS)
Implementing E-Transactions with Asynchronous Replication
IEEE Transactions on Parallel and Distributed Systems
Transaction Processing: Concepts and Techniques
Transaction Processing: Concepts and Techniques
The Database State Machine Approach
Distributed and Parallel Databases
Using Optimistic Atomic Broadcast in Transaction Processing Systems
IEEE Transactions on Knowledge and Data Engineering
Processing Transactions over Optimistic Atomic Broadcast Protocols
ICDCS '99 Proceedings of the 19th IEEE International Conference on Distributed Computing Systems
Total order broadcast and multicast algorithms: Taxonomy and survey
ACM Computing Surveys (CSUR)
Unification of Transactions and Replication in Three-Tier Architectures Based on CORBA
IEEE Transactions on Dependable and Secure Computing
Hi-index | 0.00 |
Critical database applications require 2-safe replication between at least two sites for disaster-tolerant services. At the same time, they must provide consistent and low-latency results to their clients in normal cases. In this paper, we propose Optimistic Transactional Active Replication (OTAR), which replicates the transaction logs with low latency and provides a consistent view to database applications. The latency of our replication is lower than Passive Replication, and guarantees the serializability of transaction isolation levels that cannot be supported by Active Replication. For our replication, each client sends a transaction request to all replicas and all of the replicas execute the request and optimistically return the result of the transaction to the client. Each replica generates a causality history of the transaction, sent to the client with the result. With the causality histories, the client can make sure that the requested transaction was executed in the same order at all of the replicas and eventually commit it. If the client cannot validate the order, then the client waits for the pessimistic result of the transaction from the replicas. This paper describes the algorithm and its properties.