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

Voice Service Quality Deteriorated When the Load Was Unbalanced on Three Next-Hop Links of an NE40

Publication Date:  2013-10-08 Views:  30 Downloads:  0
Issue Description
The G9 carrier network between two cities was set up using a single plane. In each city, an NE40 was deployed; 21 E1 links were configured between NE40s and each seven E1 links were bound into an MP-group (three MP-groups in total). The two NE40s ran OSPF to ensure that three equal-cost routes were always available between them. Load balancing by packet was configured. According to the traffic plan, 21 E1 links were sufficient to carry all voice and signaling traffic of the G9 network. However, the voice service quality deteriorated greatly during the peak hours starting from 20:00 each Friday.
Handling Process

To address the issue, Huawei performed the following operations and observed the following information:

1. Checked the core network side and did not find related alarms.

2. Cleared all the port statistics of NE40 and then counted packets at ports. Found the number of lost packets in an MP-group kept increasing, while the situation did not occur on the other MP-groups.

3. Confirmed that 21 E1 links were working properly and excluded the assumption that the issue was caused by decreased transmission bandwidth.

4. Viewed the traffic statistics on the NMS and found that the traffic volume in the three MP-groups kept a proportion of 1.2:1:1.

5. Confirmed that NE40 was unable to implement load balancing by packet over three next-hop links and the actual proportion of traffic over the three links was 1.2:1:1. When the transmission bandwidth of the three MP-groups was the same, one MP-group started to discard packets once the traffic volume to be transmitted was higher than 88.67% of the total bandwidth provided by the three MP-groups. 

6. Added one E1 link to the MP-group carrying traffic 1.2 times that of the other MP-groups according to the proportion. The method applies only to a fixed service model. The load balancing proportion was determined by hash algorithm based on destination IP addresses, and therefore the service model was likely to change frequently and the MP-group carrying traffic 1.2 times that of the other groups would not be determined . The method in this case did not work.

Alternatively, set the voice call limit on the UMG according to the 88% of the total bandwidth provided by the three MP-groups.
Root Cause
NE40 was unable to implement load balancing by packet over three next-hop links and the actual proportion of traffic over the three links was 1.2:1:1. When the transmission bandwidth allocated to the three MP-groups was the same, one MP-group started to discard packets once the traffic volume to be transmitted was higher than 88.67% of the total bandwidth provided by the three MP-groups.
Solution
Ran the sipxordip command to implement load balancing for Layer 3 traffic.
Suggestions
NE40 is able to implement 1:1 load balancing by packet over two next-hop links, whereas it is unable to implement load balancing by packet over three next-hop links, and the actual proportion of traffic over the three links is 1.2:1:1. The next-hop link to carry traffic 1.2 times that of the other links is determined by running the hash algorithm based on the destination IP addresses.

END