Method and device for scheduling virtual disk input and output ports

1. A method for scheduling virtual disk input and output (I/O) ports, comprising:

assigning a set of service quality ratings to a corresponding set of virtual disk I/O ports based on a set of reading-writing bandwidth quotas associated with the corresponding set of virtual disk I/O ports in a physical machine;

determining a total forecast value of a data bandwidth to be used by a plurality of reading-writing requests within a current time segment and determining virtual disk I/O ports corresponding to the plurality of reading-writing requests;

allocating reading-writing bandwidth limits to the virtual disk I/O ports corresponding to the plurality of reading-writing requests, including allocating the total forecast value of the data bandwidth to different virtual disk I/O ports according to a proportion of a service quality rating of a virtual disk I/O port of the total forecast value, wherein the allocating of the reading-writing bandwidth limits to the virtual disk I/O ports comprises:

allocating a first token container bucket to a first virtual disk I/O port, the first token container bucket comprising a plurality of token containers, and a capacity of each token container corresponding to a size of a corresponding reading-writing bandwidth limit allocated to the first virtual disk I/O port used to control scheduling of the virtual disk I/O ports; and

scheduling virtual disk I/O ports on the physical machine based at least in part on actual bandwidths to be used by the plurality of reading-writing requests within the current time segment and the reading-writing bandwidth limits for the virtual disk I/O ports corresponding to the plurality of reading-writing requests, wherein the scheduling of the virtual disk I/O ports comprises:

determining whether a number of bytes associated with the actual bandwidths to be used by the plurality of reading-writing requests for the current time segment exceeds a total capacity of the first token container bucket and

in response to a determination that the number of bytes for the current time segment exceeds the total capacity of the first token container bucket:

delaying to perform the plurality of reading-writing requests until a total capacity of the first token container bucket in a subsequent time segment can accommodate the number of bytes.

2. The method as described in claim 1 , wherein an assigned service quality rating relates proportionally to a reading-writing bandwidth quota of each virtual disk I/O port.

3. The method as described in claim 1 , wherein a sum of data bandwidth actually generated by a plurality of reading-writing requests in a previous time segment serves as the total forecast value of the data bandwidth to be used by the plurality of reading-writing requests within the current time segment.

4. The method as described in claim 1 , wherein the allocating of the reading-writing bandwidth limits comprises:

independently allocating reading bandwidth limits and writing bandwidth limits according to reading requests and writing requests respectively.

5. The method as described in claim 4 , wherein a ratio of a reading bandwidth limit and a writing bandwidth limit allocated to a corresponding I/O port corresponds to a ratio of a volume of data read from a virtual disk and a volume of data written into a virtual disk.

6. The method as described in claim 1 , wherein the capacity of each token container is proportional to the service quality ratings of the corresponding virtual disk I/O ports.

7. The method as described in claim 1 , wherein the token container buckets allocated to the different virtual disk I/O ports include a same number of token containers.

8. The method as described in claim 1 , wherein the allocating of the reading-writing bandwidth limits to the virtual disk I/O ports comprises:

adjusting in real-time distribution of token containers in the token container bucket corresponding to each virtual disk I/O port based on reading-writing requests in different time segments.

9. The method as described in claim 8 , wherein a real-time capacity of the token container bucket corresponding to each virtual disk I/O port is no greater than a quota capacity of the corresponding token container bucket.

10. The method as described in claim 9 , wherein the scheduling of the virtual disk I/O ports on the physical machine comprises:

providing token container compensation between different token container buckets of different virtual disk I/O ports.

11. The method as described in claim 1 , wherein the allocating of the reading-writing bandwidth limits to the virtual disk I/O ports comprises:

in the event that a change in quantity of virtual disk I/O ports occurs in the physical machine, re-allocating reading-writing bandwidth limits based on a total forecast value of data bandwidths of virtual disk I/O ports after the change in quantity of virtual disk I/O ports occurs.

12. The method as described in claim 1 , wherein prior to the scheduling of the virtual disk I/O ports on the physical machine, the method comprises generating weighted request queues based on correspondence of the plurality of reading-writing requests and requests for the virtual disk I/O ports to access the plurality of reading-writing requests in chronological order from the weighted request queues.

13. A device for scheduling virtual disk input and output (I/O) ports, comprising:

at least one processor configured to:

assign a set of service quality ratings to a corresponding set of virtual disk I/O ports based on a set of reading-writing bandwidth quotas associated with the corresponding set of virtual disk I/O ports in a physical machine;

determine a total forecast value of a data bandwidth to be used by a plurality of reading-writing requests within a current time segment and determine virtual disk I/O ports corresponding to the plurality of reading-writing requests;

allocate reading-writing bandwidth limits to the virtual disk I/O ports corresponding to the plurality of reading-writing requests, including allocating the total forecast value of the data bandwidth to different virtual disk I/O ports according to a proportion of a service quality rating of a virtual disk I/O port of the total forecast value, wherein the allocating of the reading-writing bandwidth limits to the virtual disk I/O ports comprises to:

allocate a first token container bucket to a first virtual disk I/O port, the first token container bucket comprising a plurality of token containers, and a capacity of each token container corresponding to a size of a corresponding reading-writing bandwidth limit allocated to the first virtual disk I/O port used to control scheduling of the virtual disk I/O ports; and

schedule virtual disk I/O ports on the physical machine based at least in part on actual bandwidths to be used by the plurality of reading-writing requests within the current time segment and the reading-writing bandwidth limits for the virtual disk I/O ports corresponding to the plurality of reading-writing requests, wherein the scheduling of the virtual disk I/O ports comprises to:

determine whether a number of bytes associated with the actual bandwidths to be used by the plurality of reading-writing requests for the current time segment exceeds a total capacity of the first token container bucket; and

in response to a determination that the number of bytes for the current time segment exceeds the total capacity of the first token container bucket:

delay performing the plurality of reading-writing requests until a total capacity of the first token container bucket in a subsequent time segment can accommodate the number of bytes; and

a memory coupled to the at least one processor and configured to provide the at least one processor with instructions.

14. The device as described in claim 13 , wherein the allocating of the reading-writing bandwidth limits comprises:

independently allocating reading bandwidth limits and writing bandwidth limits according to reading requests and writing requests respectively.

15. The device as described in claim 14 , wherein a ratio of a reading bandwidth limit and a writing bandwidth limit allocated to a corresponding I/O port corresponds to a ratio of a volume of data read from a virtual disk and a volume of data written into a virtual disk.

16. The device as described in claim 13 , wherein the allocating of the reading-writing bandwidth limits to the virtual disk I/O ports comprises:

adjusting in real-time distribution of token containers in the token container bucket corresponding to each virtual disk I/O port based on reading-writing requests in different time segments.

17. The device as described in claim 16 , wherein the scheduling of the virtual disk I/O ports on the physical machine comprises:

providing token container compensation between different token container buckets of different virtual disk I/O ports.

18. The device as described in claim 13 , wherein the allocating of the reading-writing bandwidth limits to the virtual disk I/O ports comprises:

in the event that a change in quantity of virtual disk I/O ports occurs in the physical machine, re-allocating reading-writing bandwidth limits based on a total forecast value of data bandwidths of virtual disk I/O ports after the change in quantity of virtual disk I/O ports occurs.

19. A computer program product for scheduling virtual disk input and output (I/O) ports, the computer program product being embodied in a tangible non-transitory computer readable storage medium and comprising computer instructions for:

assigning a set of service quality ratings to a corresponding set of virtual disk I/O ports based on a set of reading-writing bandwidth quotas associated with the corresponding set of virtual disk I/O ports in a physical machine;

determining a total forecast value of a data bandwidth to be used by a plurality of reading-writing requests within a current time segment and determining virtual disk I/O ports corresponding to the plurality of reading-writing requests;

allocating reading-writing bandwidth limits to the virtual disk I/O ports corresponding to the plurality of reading-writing requests, including allocating the total forecast value of the data bandwidth to different virtual disk I/O ports according to a proportion of a service quality rating of a virtual disk I/O port of the total forecast value, wherein the allocating of the reading-writing bandwidth limits to the virtual disk I/O ports comprises:

allocating a first token container bucket to a first virtual disk I/O port, the first token container bucket comprising a plurality of token containers, and a capacity of each token container corresponding to a size of a corresponding reading-writing bandwidth limit allocated to the first virtual disk I/O port used to control scheduling of the virtual disk I/O ports; and

scheduling virtual disk I/O ports on the physical machine based at least in part on actual bandwidths to be used by the plurality of reading-writing requests within the current time segment and the reading-writing bandwidth limits for the virtual disk I/O ports corresponding to the plurality of reading-writing requests wherein the scheduling of the virtual disk I/O ports comprises:

determining whether a number of bytes associated with the actual bandwidths to be used by the plurality of reading-writing requests for the current time segment exceeds a total capacity of the first token container bucket; and

in response to a determination that the number of bytes for the current time segment exceeds the total capacity of the first token container bucket:

delaying to perform the plurality of reading-writing requests until a total capacity of the first token container bucket in a subsequent time segment can accommodate the number of bytes.