ARL-Caching: A Blueprint for Mitigating Recurrent Aborts and Minimizing Computational Overhead in Transactional Systems

Abstract

Efficient cache management is essential for maintaining high performance and system integrity in transactional systems. Recurrent aborts—failures that necessitate transaction restarts—pose significant challenges, leading to increased overhead and reduced throughput. This paper introduces the ARL-Caching (Abort-Resistant and Low-Effort) blueprint, a novel approach designed to mitigate recurrent aborts while minimizing additional computational burden. The ARL-Caching blueprint presents an optimized cache management algorithm that prioritizes transactions with a lower likelihood of aborts, thereby reducing abort rates and enhancing overall system efficiency. The proposed blueprint demonstrates a significant reduction in recurrent aborts through comprehensive simulations compared to traditional cache management techniques. Our findings reveal that ARL-Caching can improve cache hit rates and system throughput while ensuring that the added effort for abort management remains minimal. The results underscore the potential of ARL-Caching to serve as a robust solution for transactional systems where aborts are frequent, providing a pathway to more reliable and efficient data processing. The implications of this work extend to various applications, including database systems, distributed computing, and other domains reliant on transactional integrity and performance.