What is DRS feature in FusionCompute?
DRS, Dynamic Resource Scheduler: dynamically allocates and balances resources, using intelligent scheduling.
According to the load condition of the system, the resources are intelligently scheduled to achieve load balancing of the system and ensure a good user experience of the system.
On the FusionCompute, you can configure the dynamic resource scheduling function of the cluster. The intelligent load balancing scheduling algorithm is used to periodically check the load of hosts in the cluster and migrate virtual machines between different hosts to achieve load balancing among hosts in the cluster. .
There are two levels of automation for a cluster: automatic and manual. In automatic scheduling mode, the system automatically migrates virtual machines to the most appropriate host. In the manual operation mode, the system generates operation suggestions for the administrator to choose, and the administrator decides whether to apply the suggestions according to the actual situation.
Resource scheduling can be configured with different metrics and can be scheduled based on CPU, memory, or CPU and memory. Virtual machine performance is only affected when resources are reused. Therefore, if memory resource reuse is not used, it is recommended to configure it to be scheduled according to the CPU. If memory reuse is used, it is recommended to configure it to be based on CPU and memory synthesis.
Advanced rules for resource scheduling can be used to meet specific needs. For example, when two virtual machines are in active/standby relationship, you can configure a mutual exclusion policy to run on different hosts to improve reliability.
The time-sharing threshold setting of resource scheduling can meet the scheduling requirements of different time periods. As the virtual machine migration brings a certain system overhead, it is recommended to set a conservative policy when the service pressure is high. If the service pressure is small, set it to a medium or aggressive policy to avoid affecting service performance.
