Scheduling virtual CPU based on priority calculated using time statistics
The present invention provides a scheduling method for a data processing system comprising at least one physical CPU, and one or more virtual machines each assigned to one or more virtual CPUs, the method comprising: a first scheduling step in which one of said virtual machines is elected to run on said physical CPU; and a second scheduling step in which at least one of the virtual CPUs assigned to the elected virtual machine is elected to run on said physical CPU. The second scheduling step is applied to the virtual machine only. When a virtual machine instance is elected to run on a given CPU, the second level scheduling determines the virtual CPU instance to run. The second level scheduling is global and can cause a virtual CPU migration from one physical CPU to another. In order to ensure correct task scheduling at guest level, virtually equivalent (in terms of calculation power) virtual CPUs should be provided to the scheduler. This is achieved by the second level scheduler using a virtual CPU election criteria based on time statistics.
1. A scheduling method for a data processing system comprising at least one physical CPU, and one or more virtual machines each assigned to one or more virtual CPUs, the method comprising:
assigning a scheduling priority P to each of said virtual CPUs in accordance with a waiting time Tw associated with each of said virtual CPUs, wherein said waiting time Tw reflects the time for transition from ready to running state and the selection of at least one of said virtual CPUs is performed based on the scheduling priority P;
a first scheduling step in which one of said virtual machines is elected to run on said physical CPU;
a second scheduling step in which at least one of the virtual CPUs assigned to the elected virtual machine is elected to run on said physical CPU, wherein the second scheduling step is performed on said physical CPU, and wherein the election is performed based on the scheduling priority P and wherein said second scheduling step comprises selecting said at least one of said virtual CPUs based on a waiting time Tw associated with each of said virtual CPUs, wherein said waiting time reflects the time for transition from ready to running state; and
selecting the virtual CPU associated with the highest scheduling priority P calculated in accordance with the following formula:
P=Tw+Bm+Bl
wherein
Bm=predetermined interrupt masking bonus
Bl=predetermined locality bonus.
2. The method of claim 1 , wherein Bm is determined to promote virtual CPUs with masked interrupts for selection in the second scheduling step.
3. The method of claim 1 , wherein Bl is determined to promote local virtual CPUs already selected to run on said physical CPU.
4. Data processing system, comprising:
at least one physical CPU;
one or more virtual machines each assigned to one or more virtual CPUs; and
a scheduler comprising:
a first scheduler for electing one of said virtual machines to run on said physical CPU; and
a second scheduler for electing at least one of the virtual CPUs assigned to the elected virtual machine to run on said physical CPU;
wherein the second scheduler is adapted to perform a scheduling operation on said physical CPU, the scheduling operation comprising selecting said at least one of said virtual CPUs based on a waiting time Tw associated with each of said virtual CPUs, wherein said waiting time reflects the time for transition from ready to running state;
wherein said data processing system is adapted to assign a scheduling priority P to each of said virtual CPUs in accordance with the waiting time Tw associated with each of said virtual CPUs, and to perform the selection of said at least one of said virtual CPUs based on the scheduling priority P;
wherein said data processing system is adapted to select the virtual CPU associated with the highest scheduling priority P calculated in accordance with the following formula:
P=Tw+Bm+Bl
wherein
Bm=predetermined interrupt masking bonus
Bl=predetermined locality bonus.
5. The data processing system of claim 4 , adapted to determine Bm to promote virtual CPUs with masked interrupts for selection in the second scheduling step.
6. The data processing system of claim 4 , adapted to determine Bl to promote local virtual CPUs already selected to run on said physical CPU.