In edge computing, it is important to ensure services running in containers on resource-constrained end devices to keep up-time parity with higher-end devices and not to fail under high loads. This research addresses this issue by experimenting with state management via migration and checkpoint creation, reducing stress caused by resource usage on any given system. The presented approach elaborates on a migration-based solution where every host in a decentralized edge system can independently manage resources of local containers and initiate their migrations as needed. To ensure minimal disruption, a container checkpoint creation and management system is built on the decentralized orchestration framework. Slow transfers of large checkpoints is tackled by way of creating patches: lightweight files made by comparing two checkpoints, which can be created non-invasively using system calls provided by the kernel. Patches created between consecutive checkpoints are utilized to realize an incremental state transfer. This method is improved with a decision making process for between 3 \(\times \) and 100 \(\times \) reduction in transfer size and latency by choosing between the patch and the checkpoint.