No relevant resource is found in the selected language.

This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>

Reminder

To have a better experience, please upgrade your IE browser.

upgrade

The LPUK on an NE40E Fails to Load Balance Traffic on MP-Groups, Causing the Voice Quality Deterioration.

Publication Date:  2013-01-07 Views:  20 Downloads:  0
Issue Description

Based on the feedback from the office in Z country, the quality of the voice service provided by the NE40E at a Site in M country goes poor and the voice service is interrupted. The voice service restores after several hours.

The software version running on the NE40E on the live network is V600R001C00SPC800.

Handling Process

The on-site Huawei technical support personnel log in to the live network and perform the ping test on the interface of the NE40E that provides the voice service. They find packet loss occurs on the interface. They check the interface status on MP-Group 2/0/1 and find that traffic is not load balanced on the three MP-Group interfaces. Therefore, they take the following preventative measures:

After consulting the core network technical engineers, the on-site Huawei technical support personnel know that all data flows carry UDP packets and the destination port numbers of the UDP packets are discrete. In this case, the traffic redirection policy can be manually configured to distribute the data flows to three links for transmission.

The on-site Huawei technical support personnel configure a traffic redirection policy in the user view on the NE40Es and use complex traffic classification to classify the destination port numbers of the UDP packets into three categories MP-Group 2/0/1, MP-Group 2/0/2, and MP-Group 2/2/1. The UDP packets are redirected to the three outbound interfaces. In this way, the data flows are load balanced manually.

Root Cause

The on-site Huawei technical support personnel check the interface status on MP-Group 2/0/1 and find that the bandwidth usage exceeds 90%. However, the traffic volume on MP-Group 2/0/2 and MP-Group 2/2/1 is much smaller than that on MP-Group 2/0/1. Therefore, based on the preliminary analysis, traffic congestion occurs on MP-Group 2/0/1, which causes the packet loss. The problem is that load balancing is configured on the three MP-Group interfaces. The traffic should not be unevenly distributed on the three interfaces.

They check the type of the LPUs installed on the NE40Es and find that the LPU on which the three MP-Groups are configured is an LPUK. An LPUK supports the hash algorithm that hashes the source and destination IP addresses. If the source and destination IP addresses are the same, the LPUK considers the traffic sent by different servers to be the same and assigns the traffic to the same link, irrespective of different port numbers and protocols. Therefore, when the network traffic contains only a few IP address samples, the link usage of one of the MP-Groups is higher than that of the other links. If this case occurs, this is caused by the hardware restriction of the LPUK. To resolve the problem, replace the LPUK with an LPUF-10 that hashes the quintuple information.

Solution

Suggest to replace the LPUK with an LPUF-10 that hashes the quintuple information.

Suggestions
In the scenario in which an LPUK is used in the upstream direction, an NE40E can perform load balancing using the hash algorithm based on IP addresses and there may be not many IP addresses available. If this case occurs, evaluate the impacts and risks that may be caused by this problem. To implement load balancing, contact Huawei technical support personnel for suggestions on how to deploy the network based on a specific scenario and service flows.

END