Hybrid deployment strategy for software updates to the manufacturing execution system layer

Complex industrial systems consist of many heterogeneous devices running different hardware and software in a connected, layer-organized environment. Since all these software instances must be updated occasionally, and since they could affect the layers under and above, the definition of deployment strategies that will reduce downtime is necessary. In previous work, we focused on identifying common problems in software update processes and concentrated on the most effective update strategies running at the lowest (Internet of Things - IoT) and highest (Enterprise Resource Planning - ERP) levels. The result was a set of recommendations and strategies that should help minimize network utilization and processing resources and make the process as energy-efficient as possible. After that, the core effort of the research is shifted toward the Manufacturing Execution System (MES) layer - the layer that brings the higher complexity, both in terms of connectivity and software complexity. Following the actual Industry 4.0 paradigm, the software in the MES layer becomes even more critical since it is expected to integrate a whole new set of responsibilities previously belonging to various levels or external solutions. To facilitate further requests, deployment strategies are reevaluated and enriched with innovative approaches such as A/B testing and the separate update service. This paper shows the possible further development of the hybrid software deployment system when applied to the multiconnected levels, such as the MES. The adaptation shows positive results regarding the network load distribution and significant effort reduction when a rollback is needed.