METHODS AND SYSTEMS FOR PROVIDING NETWORK ACCESS REDUNDANCY

摘要

In a networked multi computer environment with redundant links, network interface cards (NICs) are commonly duplicated and teamed to provide a recovery mechanism when network components fail. Embodiments of the present inventions avoid teaming of pairs of NICs and provide a computing host with redundant network connections for the computing host through a switch that is transparent to the computing host. The computing host itself; that is, its hardware and software, is relieved of the duty of network access redundancy and returned to the simple, simplex networking operating mode. The switch replaces a potentially large number of NICs and the need for computing host network access redundancy management which leads to hardware and software cost reductions and increases the robustness and reliability of the system through redundant network access.

主权项

1. A non-transitory computer-readable medium that stores computer-executable instructions that, when executed, cause one or more processors to perform operations comprising:
running a spanning tree algorithm for a network that includes a first network edge switch, a second network edge switch, and an access switch,
wherein the access switch has a first network side port coupled to the first network edge switch using a second link,wherein the access switch has a second network side port coupled to the second network edge switch using a first link, andwherein running the spanning tree algorithm is based at least on a first port path cost and a second port path cost respectively associated with the first link and the second link; based upon a result of the spanning tree algorithm, configuring one of the first and second links as a primary link and the other of the first and second links as an alternate link,
wherein only the primary link is active and enabled to carry traffic until failure of the primary link, whereupon the alternate link is activated and enabled to carry traffic; adding a selected non-zero port path cost bias to the one of the first and second port path costs associated with the alternate link such that, in an absence of a failure of the primary link or the network-edge switch to which the primary link is coupled, the port path cost associated with the primary link is lower than the port path cost associated with the alternate link; assigning a port priority to each of the primary link and the alternate link; adding a non-zero port priority bias to the port priority assigned to the alternate link; and configuring the spanning tree algorithm to select as the primary link the one of the first and second links having a lowest port priority.