Toward Effective Multi-Capacity Resource Allocation in Distributed Real-Time and Embedded Systems
Source: Vanderbilt University
Open Distributed Real-time and Embedded (DRE) systems form the core of many mission-critical domains. These DRE systems execute in environments where system operational conditions, input workload, and resource availability cannot be fully characterized a priori. DRE system characteristics, such as multiple resource constraints and significant fluctuations in resource availability and input workload, make it hard to maintain end-to-end Quality of Service (QoS). Applications in DRE systems often require multiple resources to execute properly, and need timely allocation of those resources to maintain required QoS. In open DRE systems, system resource utilization is a function of input workload and the required QoS of applications, so runtime utilization may vary significantly from estimated values.