Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

VM Mobility based on IP addresses

IP.com Disclosure Number: IPCOM000236346D
Publication Date: 2014-Apr-21
Document File: 5 page(s) / 106K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a system for intelligently moving virtual machines (VMs) from a network-constrained server to a server that has available bandwidth. The system comprises the novel features of the ability to correlate the Internet Protocol (IP) address and traffic received on the Network Interface Card (NIC), as well as the ability to make decisions on VM mobility based on network utilization.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 01 of 5

VM Mobility based on IP addresses

Network bandwidth is an increasingly important aspect in the use of virtual machines (VMs). Based on random arrival rates and random packet sizes, the response time to a user can vary. Based on networking studies, as the traffic utilization rises above approximately 70%, the response time increases exponentially. Asymptotic throughput depicts the limit of throughput, as it becomes a factor in the user response time.*

Figure 1: Throughput versus latency

Within a network of individual physical servers in a cloud environment, each link among the physical servers inside the cloud has a limited amount of network bandwidth to be able to service the end user's laptop, automobile, personal computer, or mobile device. In a virtual machine environment, the instance can occur in which the physical server has adequate processor and memory resources, but insufficient network bandwidth exists to adequately service all the virtual machines running on the physical server. One current solution is a method to randomly select virtual machines to move, and once moved, to measure the bandwidth to determine whether the remaining virtual machines have sufficient bandwidth to execute properly. Another current tool monitors traffic flow for specific applications at the application level. This has two drawbacks. It only monitors application traffic and it does not make any modifications based on this information.

A hypervisor is a software layer within the physical server that allocates and manages virtual machine operating systems within the physical server. As such, the hypervisor associates the traffic from the network to a virtual machine running within that server. In other words, the hypervisor associates the Internet Protocol (IP) address of the network traffic with a specific virtual machine. Since this association is made at the virtual machine level, rather than at the protocol level, any application network traffic is associated with the virtual machine. When the network utilization begins to affect the end user response time, a method is needed to measure the network utilization of a

1


Page 02 of 5

specific IP address/virtual machine combination and intelligently move virtual machines

to physical servers that are less utilized by the network.

The novel contribution is a system for intelligently moving VMs from a network-constrained server to a server that has available bandwidth. The system comprises the novel features of the ability to correlate the IP address and traffic received on the Network Interface Card (NIC), as well as the ability to make decisions on VM mobility based on network utilization.

To measure the network utilization of each virtual machine, it is necessary to correlate the IP address to the bytes of data received in the management information base (MIB) in the network interface card (NIC). Known in the art is the capability to tabulate the number of bytes received based on an IP address. F...