Browse Prior Art Database

A method to prevent network outage due to loops in Ethernet network

IP.com Disclosure Number: IPCOM000236974D
Publication Date: 2014-May-23
Document File: 5 page(s) / 67K

Publishing Venue

The IP.com Prior Art Database

Abstract

Broadcast messages can cause flooding and bring a network outage when a loop exists in a network. Most network administrators are familiar with the problem of network loops and are usually careful enough to avoid it. However, with the advent of virtualization, it is the server administrators who manage the switching aspects on the hypervisor. It is easy to create network loop in a virtualized environment as well. The measures to be taken to avoid network loops are the same in both the scenarios. However, these measures are manual configurations today. While physical switch administrators are much more cognizant of the network loop problem, the server administrators are not as experts as network administrators. Therefore, it happens to be a frequent event that a mis-configured server hypervisor creates a loop and therefore a broadcast flood, resulting in a network outage. Such network outages have a very serious impact since all the virtual machines running on the hypervisors connected to the impacted physical switch become unreachable.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 29% of the total text.

Page 01 of 5

A method to prevent network outage due to loops in Ethernet network
Broadcast messages can cause flooding and bring a network outage when a loop exists in a network. For example, a loop formation can be explained in the picture below, in the context of 2 physical switches connected to one another, through two links:

In Picture 1, it can be see that a broadcast packet sent by the Host through port 'a3' on 'Switch A' will keep traversing endlessly. Unlike in IP networks, there is no concept of TTL (Time-To-Live) in Ethernet switched networks, so the packet will keep traversing until it gets finally dropped when the resources are exhausted on the network switches. Picture 2 is similar to .

Most network administrators are familiar with the problem of network loops and are usually careful enough to avoid it. However, with the advent of virtualization, it is the server administrators who manage the switching aspects on the hypervisor. It is easy to create network loop in a virtualized environment as well, as can be illustrated by the following picture:

1


Page 02 of 5

It can be noticed that the Picture 3 is very similar to Picture 1, with the difference being that one of the physical switches is replaced by a hypervisor virtual switch, and the broadcast packet sender is a virtual machine on that hypervisor.

The measures to be taken to avoid network loops are the same in both the scenarios. However, these measures are manual configurations today. While physical switch administrators are much more cognizant of the network loop problem, the server administrators are not as experts as network administrators. Therefore, it happens to be a frequent event that a mis-configured server hypervisor creates a loop and therefore a broadcast flood, resulting in a network outage. Such network outages have a very serious impact since all the virtual machines running on the hypervisors connected to the impacted physical switch become unreachable.

In this invention, we propose that we create a pipeline of checks, that include all the traditional measures taken to prevent network loop condition. We also propose that the switch, regardless of whether it is a virtual switch inside the hypervisor, or the physical switch to which the hypervisor is connected, perform the sequence of checks, before allowing broadcast or multicast data traffic to go through.

The idea presented in this invention is targeted at situations where neither aggregation nor other known methods are used to prevent network loops.

The following flow chart illustrates the sequence of steps:

2


Page 03 of 5

We will describe the different measures followed today, through, manual configurations, to prevent network loops.

Traditionally, more than one links are created between two adjacent devices for the following purposes:


(a) To create a link with a higher bandwidth
(b) To provide redundancy at a port level

To create a single logical link with a bandwidth equal to the aggregate of each of the in...