Disk Scheduling for Intermediate Results of Large Join Queries in Shared-Disk Parallel Database Systems
Shared-disk database systems offer a high degree of freedom in the allocation of workload compared to shared-nothing architectures. This creates a great potential for load balancing but also introduces additional complexity into the process of query scheduling. This report surveys the problems and opportunities faced in scan processing in a shared-disk environment. We list the parameters to tune and the decisions to make, as well as some known solutions and commonsense considerations, in order to identify the most promising areas of future research.