State management in parallel systems: reframed as runtime control problem

A new paper from Shuhao Zhang and eight co-authors, submitted to ACM Computing Surveys, challenges the conventional view of state management in parallel and distributed systems. Titled "Beyond Storage: State as a Runtime Control Problem in Parallel and Distributed Systems," the 53-page literature review covers streaming, serving, retrieval, and continual-learning systems. The authors argue that shared state increasingly governs both performance and failure, but existing research isolates critical aspects like access control, hardware-aware execution, memory management, and long-horizon updates.

The review organizes the literature around three coupled dimensions: state access and scheduling, state-aware execution, and state evolution and reuse. By synthesizing across these dimensions using a common scaffold—state object, control surface, coupling path, evaluation boundary, and unresolved contract—the authors identify recurring anti-patterns. This analysis informs a contract-oriented blueprint and a disturbance-aware evaluation agenda that treats state management as a runtime control problem rather than a pure storage issue. The work is relevant for engineers building modern distributed systems that need to balance performance, consistency, and fault tolerance.