Design and Implementation of a Lightweight Distributed Container Orchestration Engine with Custom Scheduling and Fault-Tolerant Consensus

Container orchestration platforms such as Kubernetes have become foundational to cloud-native infrastructure; however, their operational complexity and resource footprint render them unsuitable for resource-constrained edge and embedded environments. This paper presents the design and implementation of a lightweight distributed container orchestration engine built in Go, targeting contexts where simplicity, low overhead, and deterministic behavior are paramount. The engine integrates three principal innovations: (1) a Predictive Scheduler that computes multi-factor composite scores across worker nodes using CPU utilization, memory availability, and current task count to achieve dynamic load balancing; (2) a Raft-based consensus module ensuring all scheduling decisions are durably committed and replicated before execution, providing strong consistency guarantees; and (3) a heartbeat-based fault detector that identifies node failures within 30 seconds and automatically reschedules affected workloads to surviving nodes. Experimental evaluation on a three-node cluster demonstrates correct task distribution, load-balancing penalty mechanics, and transparent fault recovery with near-zero rescheduling latency. The results confirm that coordinated consensus and predictive scheduling can be unified in a compact Go runtime without sacrificing fault tolerance or operational simplicity.