Hi, there has been a heated discussion about dynamic scheduling last week.(http://www.mail-archive.com/[email protected]/msg21644.html) I am also interested in this topic. We believe that dynamic scheduling consists of two parts: balancing computing capacity and optimizing power consumption. For balancing computing capacity, the ceilometer periodically monitors distribution and usage of CPU and memory resources for hosts and virtual machines. Based on the information, the scheduler calculates the current system standard deviation metric and determines the system imbalance by comparing it to the target. To resolve the imbalance, the scheduler gives the suitable virtual machine migration suggestions to nova. In this way, the dynamic scheduling achieves higher consolidation ratios and deliver optimized performance for the virtual machines. For optimizing power consumption, we attempt to keep the resource utilization of each host within a specified target range. The scheduler evaluates if the goal can be reached by balancing the system workloads. If the resource utilization of a host remains below the target, the scheduler calls nova to power off some hosts. Conversely the scheduler powers on hosts to absorb the additional workloads. Thus optimizing power consumption offers an optimum mix of resource availability and power savings. As Chen CH Ji said, "nova is a cloud solution that aim to control virtual / real machine lifecycle management the dynamic scheduling mechanism is something like optimization of the cloud resource". We think implementing the dynamic scheduling with heat may be a good attempt. Do you have any comments? Thanks, Jenny
_______________________________________________ OpenStack-dev mailing list [email protected] http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
