Relaxing consistency requirement for read-only transactions

In this paper, we investigate the approach to relaxing consistency requirement of serializability to improve the performance of real-time database systems. We make use of the property that a read-only transaction (ROT) only reads, but does not update any data item. Since there is a significant proportion of ROTs in several real-time systems, it is important to investigate how to process ROTs effectively. We define a weaker form of consistency, view consistency, which allows ROTs to perceive different serialization orders of update transactions. Although view consistency permits non-serializablility, ROTs are still ensured to see consistent data. We devise a separate algorithm to process ROTs from update transactions for view consistency. For comparison, we devise another separate algorithm to process ROTs for strong consistency that is equivalent to serializability. Our simulation results show that the two separate algorithms outperform the locking and optimistic protocols significantly. Using separate algorithms reduces the interference between ROTs and update transactions. This reduced interference alleviates the impact of concurrency control on real-time priority-driven scheduling and improves the timeliness of the system. To our great surprise, the two separate algorithms perform almost the same. In our investigation, we found that most data conflicts between an ROT and update transactions are independent data conflicts. These independent data conflicts can be easily resolved by using a separate algorithm even under serializability. Thus, this observation indicates that using a separate algorithm to process ROTs from update transactions is more effective than relaxing consistency requirement.