Automated Redeployment of Real-Time Systems Informed by User-Provided Workflows

Distributed, real-time and embedded (DRE) systems are a subset of computing applications that require stringent quality-of-service despite operating in environments with scarce CPU, memory, or network resources. A general lack of automated tools for redeploying DRE systems in a timely manner upon incurring failures or degradation in QoS have required an engineer or developer to reboot the system and redeploy the important components manually. Whether this type of delay is acceptable to users of the DRE system depends on the context of the application usage, but for mission-critical applications, such outages and delays can result in loss of money and lives. In this paper, we discuss our efforts and elaborate on the challenges in the area of middleware to faciliate dynamic redeployment of continuously available and potentially mission-critical DRE systems. Keywords-redeployment, constraint satisfaction, optimization, quality of service, cyberphysical systems