Checkpointing Methods and Their Effectiveness in Keyboard Data Entry Operations with Large Access Overhead

Checkpoints are the most crucial component of the failure recovery function, established to set the starting point of recovery operations and to limit their quantity. In this study, we explore the effectiveness of checkpointing methods, i.e., checkpoint creation methods in keyboard data entry operations with extremely high access overheads. The most significant feature of checkpointing here is that the overhead for data writing depends on the accumulated amount of input data in primary memory, leading to increased checkpointing overhead and reduced operational efficiency as input operations progress. We evaluate and compare the effectiveness of six checkpointing methods to address this issue. The evaluation model is described by a simple probabilistic model, and the total expected operational time is derived as an evaluation measure. Moreover, in the numerical analysis, results from a small-scale survey and actual measurements are applied to model parameters such as the failure occurrence rate for various considerations. These results show that effective checkpointing is possible with any of the methods, especially the method of strategically splitting the file for input to reduce checkpointing overhead is superior.