Optimal Checking Times of Backup Operation for a Database System

Abstract

AbstractA database system works for tasks with random processing times. Checking times are generated at every completion of the Kth processing time. When a failure occurs in the process of the system, we execute the backup operation until the latest checking time and make the recovery of database files. The expected cost is obtained by using an inspection policy, and when failure and random working times are exponential, an optimal policy which minimizes it is derived analytically. As a special case of each task with constant time, this is compared numerically with that of random time. Furthermore, as one example of a job with a finite number of tasks, the optimization problem in what place we set checking points for a job with six tasks is considered.KeywordsRandom taskChecking timeBackup operationPeriodic timeSix tasks