A class of concurrency control algorithms to improve the schedulability of a transaction system for real-time databases is proposed. The protocols are based on the idea of aborting transactions that introduce excessive blocking. The impact of aborting a relationship is evaluated when the relationship is built in an online or offline fashion. The overhead of aborting transactions is measured on a system running a real-time operating system. Effectiveness of the proposed algorithms is demonstrated through some case studies.
The paper constitutes a useful contribution to the field of scheduling transactions in real-time database systems. It should be accessible to readers with a general background in real-time scheduling. The list of citations is also helpful for researchers working in the field.