Abstract: In distributed computing systems, the continuous execution of services places increasing demands on the ability to perform incremental state recognition, consistent synchronization across nodes, and seamless runtime switching. Existing mechanisms often lack fine-grained structural awareness, coordinated consistency control, and controllable version transitions, making it difficult to support stable service evolution under frequent state changes.This paper proposes Zynco, a lightweight runtime state coordination framework designed for state-driven distributed services. Zynco features three core mechanisms: (1) a fingerprint-based and structure-aware incremental identification method that transmits only effective state changes and maintains a sliding version chain; (2) a hybrid feedback protocol combining UDP multicast and TCP control channels to ensure efficient and reliable multi-node state distribution; and (3) a three-layer service structure with Listener-based routing to enable smooth version switching and consistent rollback. Experiments show that Zynco synchronizes 54GB of state across 20 nodes in just 27 seconds, with zero service interruption and no state tearing during switching. Under a 1000 QPS workload, response jitter remains below 8ms. Zynco provides efficient, stable, and controllable runtime support for state-intensive distributed services and demonstrates strong scalability.

Key words: Keywords: Distributed computing, Runtime state management, Incremental recognition, Multi-node synchronization, Seamless service switching

摘要： 分布式计算系统对任务持续运行过程中的运行时状态增量识别、一致同步与服务无中断切换提出了更高要求，现有机制在集群状态结构感知、节点一致性控制与服务版本切换可控性方面仍存在不足，难以支撑状态频繁演化下的稳定服务运行。为此提出轻量级运行时状态协同更新框架Zynco，包括三项核心特性：基于指纹与结构感知的增量识别机制，仅传递有效变更内容，并构建滑动版本链；结合UDP组播与TCP控制通道的反馈协议，实现状态在多节点间的高效分发与确认；借助监听器路由与三层服务结构，实现服务的平滑切换与一致性回退。实验在20节点集群完成54GB数据状态同步仅需27秒，服务切换期间任务不中断且状态无撕裂，在1000 QPS压力下性能抖动低于8毫秒。Zynco表现出的高效、稳定、可控的特性，可为海量数据协同更新场景的状态驱动型分布式高性能服务提供良好的运行支撑。

关键词: 布式计算系统, 运行时状态, 增量识别, 多节点一致性同步, 服务无缝热切换