A Latency-Tolerant Extension of the Raft Algorithm for Read-Only Operations in Spaceborne Systems

With the continuous advancement of satellite technology and the increased deployment of satellite-borne computing systems, distributed consensus algorithms encounter significant performance degradation due to high latency and constrained communication conditions inherent in satellite communication environments. This paper addresses the performance bottleneck of the Raft algorithm specifically in processing read-only requests under such conditions by proposing a latency-tolerant optimization scheme. Our proposed method introduces a configurable parameter termed "delay-limit", enabling clients to tolerate an acceptable degree of data staleness in read-only scenarios. Consequently, clients can directly obtain data from proximal follower nodes, thereby effectively reducing communication overhead and response latency. Building upon the consistency semantics of the original Raft algorithm, we redesign the system architecture to support latency-tolerant reads and implement the optimized Raft protocol using Go. Experimental evaluations conducted within a simulated high-latency satellite communication environment demonstrate that the proposed scheme effectively reduces the average read latency—for example, latency decreases from approximately 9500 ms to 1500 ms under conditions of 1000 ms propagation delay—and improves overall throughput by over 10%, with negligible impact on write performance. The experimental results underscore the practical applicability and adaptability of our optimization scheme in satellite-borne distributed systems, providing valuable insights for further research and engineering applications in resource-constrained, high-latency environments.