Problem Description

An overseas bureau is going to use the SSN1EAS2 board of OSN3500 to interface with the 10GELAN port of our router NE40E-X3 . The SSN3EAS2 board is used to create a load sharing lag, when sending a specific format message, the service is not shared to all ports, but to some ports, resulting in service packet loss.

The networking mode is: SSN3EAS2 NE40E-X3 10GELAN Smartbit

Alarm messages

None

Processing

There are two options to solve this problem.

Option 1

This solution is to dock the DigitalComm NE40E with the SSN1EAS2 veneer of the transport network MSTP device, in which the port types of the two devices that are docked are both in 10GE LAN mode, which is a 10GE Ethernet port.

The NE40E does not need to do other redundant operations.

SSN1EAS2 needs to complete:

1. Configure the bidirectional link between port 1 and port 2;

2、Configure port 2 as 10GE WAN mode (reflected on the network management as working mode: 10gfullwan );

3、SSN4SL64 (supports 64VC4 line boards, not limited to SL64 ) Configure 64C adjacent cascade crossover.

[Advantages] For digital communication equipment does not support the WAN mode, to solve the 10GE single stream through the 10GE

[Disadvantages]

1, waste of resources, SSN1EAS2 single board only do the work of interface conversion, did not play its role of Ethernet Over SDH;

2, MSTP equipment SL64 single board configured with 64C adjacent cascade crossover, it means that when up and down services, 10GE services can only be transmitted within a channel, the node of transmission can no longer divide the service into multiple streams.

Program 2

Pre-condition: the digital communication equipment supports 10GE WAN/LAN mode switching, and currently the NE40E supports this function

This scenario is to interface the Digicom NE40 with the SL64 line board of the transport network MSTP equipment, in which the port types of NE40 are all 10GE WAN mode

NE40E needs to complete:

1. Modify the port mode to 10GE WAN mode;

SSN4SL64 need to complete:

1. SL64 (line boards supporting 64VC4s, not limited to SL64 ) configure 64C adjacent cascade crossover.

[Advantages] 1, to solve the 10GE single stream pass-through 10GE

2、Save the budget of data single board

[Disadvantages]

OSN3500 MSTP equipment SL64 single board configured with 64C adjacent cascade crossover means that when up and down services, 10GE services can only be transmitted in one channel, and the node of transmission can no longer divide the service into multiple streams.

Root Cause

In the application environment, the device interfacing with the SSN1EAS2 single board is an MPLS message router. Therefore, the messages received by the SSN1EAS2 single board are MPLS messages. Because the single board does not recognize MPLS messages, the input to the hash algorithm after configuring the lag is the DA, SA, and other information of the MPLS message, not the DA, SA, and other information sent by the SmartBit instrument. At this time, the hash input is the router L2 header and MPLS label. hash According to the intercepted message information, there are three types of messages:

Message 1:

00e0fcad a1a100e0 fc7b35c2 88471000

42ff23df 63ff0018 821a6136 00e0fc85

8c788100 00190800 4500006e 53ff0000

3ffda420 c0240102 c04b0102 00000000

00000000 00000000 00000000 00000000 00000000 00000000

00000000 00000000 00000000

Message 2:

00e0fcad a1a100e0 fc7b35c2 88471000

40ff23df 81ff0000 5e000186 00e0fc85

8c788100 040a0800 4500006e 43160000

3ffdb508 c0240102 c04b0102 00000000

00000000 00000000 00000000 00000000

Message 3:

00188203 178f00e0 fcada1a1 88470040

80ff1280 81ff4500 006e2de5 00003efd

cb3ac024 0102c04b 01020000 00000000 00000000

00000000 00000000 00000000

As you can see, the chip calculates the same hash value for message 1 and message 2 because they have the same L2 header. The hash value of message 3 needs to be actually calculated. It is verified that message 1 and message 2 are hashed to one port and message 3 is hashed to another port.

Therefore, on the SSN3EAS2 single board, after configuring load sharing lag, the specific service sharing result is completely determined by the L2 header of the message. If, during testing, you find that the load sharing does not achieve the expected results, you need to confirm whether the L2 header of the message meets the chip's hash requirements.

Recommendations and Summary

It is recommended that when considering network configuration, you should first familiarize yourself with the indicators, application scenarios and application restrictions of the equipment boards on both sides of the docking.