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

Per-Session Load Balancing Is Configured on an NE40E for IP Trunk, Causing the GPRS Traffic to Be Congested.

Publication Date:  2013-01-08 Views:  24 Downloads:  0
Issue Description

Version: NE40E V300R003C02B697

Networking:

Internet--SW1--FW1--NE40E-A1--7×STM1(IP-TRUNK)--NE80E-B1---NE40E-B1--FW2--GGSN

                                                  |

                                                 SCG

Symptom: Based on the customer feedback, the GPRS services are transmitted slowly, causing the online users to fail to visit the pages at the expected rate.

Handling Process

The GGSN services get access to the Internet through an SCG. The GGSN pings the IP address of the SCG, and a delay and packet loss occurs during the ping process.

The GGSN pings NE40E-A1, and no delay or packet loss occurs. Based on the preliminary analysis, the problem occurs on the link between NE40E-A1 and the SCG.

The SCG is connected to NE40E-A1 through a switch. The on-site Huawei technical support personnel log in to the switch. They use the switch to ping NE40E-A1, and no delay or packet loss occurs. Then they use the switch to ping the SCG, and no problem occurs either.

They use the SCG to ping NE40E-A1, and no fault occurs.

They collect traffic statistics on NE40E-A1, NE80E-B1, and NE40E-B1 respectively. They find that the problem occurs on the link between NE80E-B1 and NE40E-A1.

They check the traffic volume on the links between NE80E-B1 and NE40E-A1 respectively using the display interface brief command. They find that the traffic volume on one of the seven member interfaces in the IP-Trunk interface from NE40E-A1 to NE80E-B1 reaches 97%, which is a symptom of traffic congestion. However, the traffic volume on the other six member interfaces is only 40% to 50%.

They doubt that the problem is caused by the hash algorithm. After consulting the Huawei GGSN engineers for the GPRS traffic model, they learn that a GRE tunnel is established between the GGSN and the SCG. All GPRS traffic is forwarded through the GRE tunnel. However, the GPRS traffic is frequently expanded recently, reaching 155 Mbit (STM-1).

They check the IP-Trunk configuration on the NE40E and NE80E. By default, the NE40E and NE80E perform per-session load balancing. The GPRS traffic is forwarded through the GRE tunnel. The NE40E hashes the source and destination IP addresses and forwards the traffic to one of the member interfaces, causing the traffic congestion.

Root Cause

By default, the NE40E and NE80E perform per-session load balancing. The GPRS traffic is forwarded through the GRE tunnel. The NE40E hashes the source and destination IP addresses and forwards the traffic to one of the member interfaces, causing the traffic congestion.

Solution

Change the load balancing mode on the NE40E for the IP-Trunk to per-packet load balancing using the load-balance packet-all command. This configuration enables the GPRS traffic to be distributed to the other member interfaces. As a result, the traffic congestion is resolved and the users can visit pages at the expected rate.

END