A Modular Abstract Framework for Devising Optimized RSMs Based on Load Distribution

Replicated state machine (RSM) is usually used for building highly available services in today's cloud systems. By analyzing typical instances and summing up their common characteristics, we introduce a modular abstract framework for devising optimized RSMs from the load distribution perspective, which is an active exploration to find a new optimization design method. The framework provides a faithful deconstruction of a class of RSMs optimized by load distribution. Through the modularization of functionalities/mechanisms, the framework decouples the design of components of a RSM to some extent and facilitates the reusability of existing and optimized functional components. By abstracting the configuration and specifying the execution procedure, the abstract skeletal protocol in this framework simplifies the description of optimized RSMs. We finally present the reconstruction of S-RSM, a RSM that is optimized based on load distribution, to illustrate the effectiveness of our framework.