Effects of distributed database modeling on evaluation of transaction rollbacks

The effect of the distributed database transaction models on the computational complexity and accuracy of the rollback statistics in a partitioned database is investigated. The author develops six probabilistic models and gives expressions for the number of rollbacks under each of these models. Essentially, the models differ in terms of the available system information. The analytical results are compared to results from simulation. It is concluded that most of the probabilistic models yield overly conservative estimates of the number of rollbacks. The effect of transaction commutativity on system throughput is also grossly undermined when such models are used.<<ETX>>