Multi-level Container Checkpoint Performance Optimization Strategy in SDDC

In order to improve the performance of large-scale cloud applications deployed on containers in Software Defined Data Centers (SDDC), this paper proposes a multi-level container checkpoint performance optimization strategy. Traditional checkpoint method exist the problem of frequent copying and transmission of dirty pages,so a based on the delayed copy incremental checkpoint algorithm predicted by dirty page degrees is proposed, reducing the size of checkpoint files from the source and solving the problem of long downtime. The goal of performance optimization is to reduce application downtime and recovery time. The effectiveness of the strategy is verified by plenty of experiments. The given experiments show that the strategy can reduce the downtime by 21%, the recovery time by 18% and the checkpoint files size by 24%.