Resource Optimization and Monitoring in Virtualized Infrastructure

• US 20110099267A1
• Filed: 10/27/2009
• Published: 04/28/2011
• Est. Priority Date: 10/27/2009
• Status: Active Grant

• Current Assignee: Vmware LLC (Broadcom, Inc.)
• Original Assignee: VMware, Inc. (Broadcom, Inc.)
• Inventors: CHILKOTI, Harish, SHRIVASTAVA, Anand, SURI, Salil
• Granted Patent: US 8,924,534 B2
• Time in Patent Office: Days
• US Class Current: 709/224
• CPC Class Codes: G06F 2009/45575 Starting, stopping, suspending or resuming virtual machine instances
• CPC Class Codes: G06F 2209/504 Resource capping
• CPC Class Codes: G06F 9/4856 resumption being on a different machine, e.g. task migration, virtual machine migration G06F9/5088 takes precedence
• CPC Class Codes: G06F 9/5022 Mechanisms to release resources
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A method includes monitoring a resource consumption of one or more virtual machines on a host computer in a computer network, and suspending the one or more virtual machines based on the resource consumption to free one or more resources on the host computer. The method also includes losslessly handling a request intended for the suspended one or more virtual machines to seamlessly resume the suspended one or more virtual machines. The request includes a network traffic intended for the suspended one or more virtual machines and/or a non-network request.

Claims

1. A method comprising:

monitoring a resource consumption of at least one virtual machine on a host computer in a computer network;

suspending the at least one virtual machine based on the resource consumption to free at least one resource on the host computer; and

losslessly handling a request intended for the suspended at least one virtual machine to seamlessly resume the suspended at least one virtual machine, the request including at least one of a network traffic intended for the suspended at least one virtual machine and a non-network request.

2. The method of claim 1, comprising monitoring at least one of a central processor utilization, a memory utilization, a network Input/Output (I/O), and a storage I/O of the at least one virtual machine.

3. The method of claim 1, wherein the non-network request includes at least one of a storage protocol request, a network protocol request, a communication between a plurality of virtual machines on the host computer, and a communication between the at least one virtual machine and the host computer.

4. The method of claim 1, further comprising profiling the resource consumption of the at least one virtual machine based on a threshold minimum of utilization of the at least one resource.

5. The method of claim 1, further comprising saving a current configuration of the at least one virtual machine prior to suspending the at least one virtual machine.

6. The method of claim 1, wherein suspending the at least one virtual machine based on the resource consumption to free the at least one resource on the host computer comprises:

decoupling the at least one virtual machine from a virtual switch to which the at least one virtual machine is coupled;

coupling the at least one virtual machine to a virtual switch having no uplink to which an appliance virtual machine is coupled; and

updating a routing information of the appliance virtual machine to enable routing of the request destined for the at least one virtual machine through the appliance virtual machine,

wherein each virtual switch is configured to enable a virtual machine coupled to the virtual switch to send and receive network traffic and to receive a communication from a kernel module.

7. The method of claim 1, comprising powering on the suspended at least one virtual machine through a user to resume the suspended at least one virtual machine.

8. The method of claim 1, further comprising permanently freeing the at least one resource on the host computer based on a threshold value of a time for which the at least one virtual machine is suspended.

9. The method of claim 6, further comprising migrating the at least one virtual machine to a host computer on which the appliance virtual machine is resident prior to coupling the at least one virtual machine to the virtual switch having no uplink to which the appliance virtual machine is coupled.

10. The method of claim 6, wherein losslessly handling at least one of the network traffic intended for the suspended at least one virtual machine and the non-network request intended for the suspended at least one virtual machine to seamlessly resume the suspended at least one virtual machine includes a corresponding at least one of:

saving packets in the network traffic destined for the suspended at least one virtual machine through the appliance virtual machine;

transmitting a network message through the appliance virtual machine to wake up the suspended at least one virtual machine; and

transmitting the packets in the network traffic destined for the suspended at least one virtual machine to the woken up at least one virtual machine

as losslessly handling the network traffic intended for the suspended at least one virtual machine, and

communicating the non-network request to the suspended at least one virtual machine through the kernel module

as losslessly handing the non-network request intended for the suspended at least one virtual machine.

11. The method of claim 8, further comprising powering on a new virtual machine on the host computer.

12. The method of claim 8, further comprising limiting the release of the at least one resource to power on a limited plurality of virtual machines.

13. The method of claim 8, further comprising preventing a predetermined allocation of the at least one resource to the at least one virtual machine.

14. The method of claim 8, further comprising:

monitoring the freed up at least one resource; and

imposing a resource limit on a new virtual machine to be powered on based on the monitoring of the freed up at least one resource.

15. The method of claim 8, further comprising optimizing the suspension of the at least one virtual machine to render the at least one virtual machine active solely during a time when the computer network experiences a peak load.

16. The method of claim 10, further comprising:

decoupling the woken up at least one virtual machine from the virtual switch having no uplink to which the appliance virtual machine is coupled; and

coupling the woken up at least one virtual machine to one of the original virtual switch to which the woken up at least one virtual machine was coupled prior to suspension and a new virtual switch,

prior to transmitting the packets in the network traffic to the woken up at least one virtual machine.

17. The method of claim 16, further comprising migrating the woken up at least one virtual machine to one of the original host computer on which the woken up at least one virtual machine was resident prior to suspension and a new host computer prior to transmitting the packets in the network traffic to the woken up at least one virtual machine based on at least one of a resource requirement of the woken up at least one virtual machine and a resource utilization of the host computer to which the woken up at least one virtual machine is moved.

18. A method comprising:

suspending at least one virtual machine on a host computer in a computer network based on a resource consumption to free at least one resource on the host computer;

queuing a request intended for the suspended at least one virtual machine at an appliance virtual machine;

sending a broadcast packet data from the appliance virtual machine to seamlessly resume the suspended at least one virtual machine; and

losslessly routing the request intended for the suspended at least one virtual machine to the resumed at least one virtual machine based on the queue at the appliance virtual machine.

19. The method of claim 18, wherein the request includes at least one of a network traffic intended for the suspended at least one virtual machine and a non-network request including at least one of a storage protocol request, a network protocol request, a communication between a plurality of virtual machines on the host computer, and a communication between the at least one virtual machine and the host computer.

20. The method of claim 19, wherein losslessly routing the non-network request intended for the suspended at least one virtual machine includes communicating the non-network request to the suspended at least one virtual machine through a kernel module.

21. A computer network comprising:

a cluster of a plurality of host computers, each of the plurality of host computers comprising at least one virtual machine;

a kernel module to monitor a resource consumption of the at least one virtual machine on a corresponding host computer; and

an appliance virtual machine coupled to a virtual switch in the kernel module and configured to enable routing of a request intended for at least one virtual machine,

wherein the at least one virtual machine is suspended by the kernel module when the resource consumption of the at least one virtual machine is less than a threshold value,

wherein a request intended for the suspended at least one virtual machine is queued at the appliance virtual machine,

wherein a broadcast packet data is sent from the appliance virtual machine to seamlessly resume the suspended at least one virtual machine, and

wherein the request intended for the suspended at least one virtual machine is losslessly routed through the appliance virtual machine to the resumed at least one virtual machine based on the queue at the appliance virtual machine.

22. The computer network of claim 21, wherein:

the at least one virtual machine is decoupled from a virtual switch thereof,

the at least one virtual machine is coupled to the virtual switch having no uplink to which the appliance virtual machine is coupled, and

a routing information of the appliance virtual machine is updated to enable routing of the request intended for the at least one virtual machine through the appliance virtual machine,

prior to the suspension of the at least one virtual machine based on the resource consumption.