Heterogeneous Cloud Systems and Criteria for Enhanced Performance

Cloud systems are managed on the basis of au-tonomous systems. We present criteria that are suitable for optimization or improvement of modern cloud systems. In every operating system, task scheduling is very important. In clouds systems, where large amount of tasks runs on numerous machines, optimized task scheduling leads to significant reduction of computing time. Cloud providers have to comply with service level agreement from technical and from the quality of service point of view. For this reason, we specify quality of service criteria and limits for service level agreement violation. Clouds are virtualized by virtual machines or containers. We show approaches for power consumption minimization. Fog computing helps to improve middleware technology between cloud computing and the IoT devices. We specify criteria for correct decomposition of parallel and distributed application. We conclude that understanding of effective work of the system, can improve design and implementation of parallel and distributed application.