Adaptive Scheduling of Data Flows in Data Center Networks for Efficient Resource Utilization

• US 20150124608A1
• Filed: 11/05/2013
• Published: 05/07/2015
• Est. Priority Date: 11/05/2013
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Agarwal, Kanak B., Carter, John B., Dixon, Colin K.
• Granted Patent: US 9,634,938 B2
• Time in Patent Office: Days
• US Class Current: 370/235
• CPC Class Codes: H04L 45/38 Flow based routing
• CPC Class Codes: H04L 45/66 Layer 2 routing, e.g. in Ethernet based MAN's
• CPC Class Codes: H04L 47/122 by diverting traffic away from congested entities

Abstract

An approach is provided in which a first virtual machine, executing on a host computer system, generates a data packet with a target destination at a second virtual machine over a computer network. The host computer system identifies a data flow corresponding to the data packet based the data packet'"'"'s header information, and analyzes path weightings of available paths that are made available to the identified data flow. In turn, the host computer system assigns one of the available paths to the identified data flow corresponding to a pre-defined physical layer path from the first virtual machine to the second virtual machine.

Claims

1. A method comprising:

identifying, at a host computer system, a data flow corresponding to a data packet based upon selected header information included in the data packet, the data packet initiated by a first virtual machine with a destination at a second virtual machine;

analyzing, at the host computer system, a plurality of path weightings of a plurality of available paths that are assignable to the identified data flow; and

assigning one of the plurality of available paths to the identified data flow based upon the analysis, wherein the assigned path corresponds to a pre-defined physical layer path from the first virtual machine to the second virtual machine over a computer network.

2. The method of claim 1 wherein the host computer system includes a virtual switch, the method further comprising:

creating, by a control plane included in the virtual switch, a flow table entry that includes a flow identifier and a path identifier, wherein the flow identifier corresponds to the selected header information and the path identifier corresponds to the assigned path;

modifying, by a data plane included in the virtual switch, the data packet by replacing a MAC destination address in a header field of the data packet with a virtual MAC address corresponding to the path identifier; and

sending the modified data packet over the computer network to the destination virtual machine.

3. The method of claim 2 wherein the flow identifier includes the selected header information comprising a source IP address, a destination IP address, a source port number, a destination port number, and a protocol identifier.

4. The method of claim 2 wherein the data packet is included in a plurality of data packets that correspond to the identified data flow, the method further comprising:

receiving a different data packet at the virtual switch, the different data packet included in the plurality of data packets;

determining, by the data plane, that different header information included in the different data packet matches the flow identifier in the flow table entry;

modifying the different data packet, by the data plane, by replacing a different MAC destination address in a different header field of the different data packet with the virtual MAC address corresponding to the path identifier; and

sending the modified different data packet over the computer network to the destination virtual machine.

5. The method of claim 1 wherein a centralized flow scheduler provides the plurality of path weightings to the host computer system, the method further comprising:

monitoring, by the centralized flow scheduler, a plurality of link utilization rates between a plurality of network devices included in the computer network;

detecting, by the centralized flow scheduler, at least one of the plurality of link utilization rates exceeds a threshold;

identifying, by the centralized flow scheduler, one or more network paths that include a link corresponding to the detected link utilization rate; and

creating, by the centralized flow scheduler, an updated path weighting for each of the one or more identified network paths.

6. The method of claim 5 wherein, for each of the one or more identified network paths, the method further comprises:

determining, by the centralized flow scheduler, a number of links included in the identified network path;

determining, by the centralized flow scheduler, the number of congested links included in the identified path; and

computing, by the centralized flow scheduler, the updated path weighting based upon the number of links and the number of congested links.

7. The method of claim 5 further comprising:

identifying, by the centralized flow scheduler, one or more host computer systems, which includes the host computer system, that utilize one or more of the identified network paths; and

sending, by the centralized flow scheduler, the one or more updated path weightings to the one or more identified host computer systems, wherein at least one of the plurality of available paths matches at least one of the one or more identified network paths.

8. The method of claim 1 wherein the computer network is a software defined data center network.

9. An information handling system comprising:

one or more processors;

a memory coupled to at least one of the processors;

a virtual switch accessible to at least one of the processors; and

a set of computer program instructions stored in the memory and executed by at least one of the processors in order to perform actions of;

identifying a data flow corresponding to a data packet based upon selected header information included in the data packet, the data packet initiated by a first virtual machine with a destination at a second virtual machine;

analyzing a plurality of path weightings of a plurality of available paths that are assignable to the identified data flow; and

assigning one of the plurality of available paths to the identified data flow based upon the analysis, wherein the assigned path corresponds to a pre-defined physical layer path from the first virtual machine to the second virtual machine over a computer network.

10. The information handling system of claim 9 wherein the virtual switch performs additional actions comprising:

creating, by a control plane included in the virtual switch, a flow table entry that includes a flow identifier and a path identifier, wherein the flow identifier corresponds to the selected header information and the path identifier corresponds to the assigned path;

modifying, by a data plane included in the virtual switch, the data packet by replacing a MAC destination address in a header field of the data packet with a virtual MAC address corresponding to the path identifier; and

sending the modified data packet over the computer network to the destination virtual machine.

11. The information handling system of claim 10 wherein the flow identifier includes the selected header information comprising a source IP address, a destination IP address, a source port number, a destination port number, and a protocol identifier.

12. The information handling system of claim 10 wherein the data packet is included in a plurality of data packets that correspond to the identified data flow, the virtual switch performing additional actions comprising:

receiving a different data packet, the different data packet included in the plurality of data packets;

determining, by the data plane, that different header information included in the different data packet matches the flow identifier in the flow table entry;

modifying the different data packet, by the data plane, by replacing a different MAC destination address in a different header field of the different data packet with the virtual MAC address corresponding to the path identifier; and

sending the modified different data packet over the computer network to the destination virtual machine.

13. The information handling system of claim 9 wherein a centralized flow scheduler provides the plurality of path weightings to the information handling system, the centralized flow scheduler performing additional actions comprising:

monitoring a plurality of link utilization rates between a plurality of network devices included in the computer network;

detecting at least one of the plurality of link utilization rates exceeds a threshold;

identifying one or more network paths that include a link corresponding to the detected link utilization rate; and

creating an updated path weighting for each of the one or more identified network paths.

14. The information handling system of claim 13 wherein, for each of the one or more identified network paths, the centralized flow scheduler performing additional actions comprising:

determining a number of links included in the identified network path;

determining the number of congested links included in the identified path; and

computing the updated path weighting based upon the number of links and the number of congested links.

15. The information handling system of claim 13 wherein the centralized flow scheduler performs additional actions comprising:

identifying one or more host computer systems, which includes the information handling system, that utilize one or more of the identified network paths; and

sending the one or more updated path weightings to the one or more identified host computer systems, wherein at least one of the plurality of available paths matches at least one of the one or more identified network paths.

16. A computer program product stored in a computer readable storage medium, comprising computer program code that, when executed by an information handling system, causes the information handling system to perform actions comprising:

identifying a data flow corresponding to a data packet based upon selected header information included in the data packet, the data packet initiated by a first virtual machine with a destination at a second virtual machine;

analyzing a plurality of path weightings of a plurality of available paths that are assignable to the identified data flow; and

assigning one of the plurality of available paths to the identified data flow based upon the analysis, wherein the assigned path corresponds to a pre-defined physical layer path from the first virtual machine to the second virtual machine over a computer network.

17. The computer program product of claim 16 wherein the information handling system includes a virtual switch, and wherein the computer program code, when executed by the information handling system, causes the information handling system to perform additional actions comprising:

creating, by a control plane included in the virtual switch, a flow table entry that includes a flow identifier and a path identifier, wherein the flow identifier corresponds to the selected header information and the path identifier corresponds to the assigned path;

modifying, by a data plane included in the virtual switch, the data packet by replacing a MAC destination address in a header field of the data packet with a virtual MAC address corresponding to the path identifier; and

sending the modified data packet over the computer network to the destination virtual machine.

18. The computer program product of claim 17 wherein the flow identifier includes the selected header information comprising a source IP address, a destination IP address, a source port number, a destination port number, and a protocol identifier.

19. The computer program product of claim 17 wherein the data packet is included in a plurality of data packets that correspond to the identified data flow, and wherein the computer program code, when executed by the information handling system, causes the information handling system to perform additional actions comprising:

receiving a different data packet at the virtual switch, the different data packet included in the plurality of data packets;

determining, by the data plane, that different header information included in the different data packet matches the flow identifier in the flow table entry;

modifying the different data packet, by the data plane, by replacing a different MAC destination address in a different header field of the different data packet with the virtual MAC address corresponding to the path identifier; and

sending the modified different data packet over the computer network to the destination virtual machine.

20. The computer program product of claim 16 wherein the computer network is a software defined data center network.