Cisco CCNP and CCIE Study Notes

If you were turn on Link Fragmentation and Interleaving (LFI) using the ppp multilink interleave in the multilink PPP example given in the last post, there would be a problem with the ordering of packets that get interleaved and sent over the pair of PPP links that are part of the multilink bundle. The reason is that the small packets that would get interleaved between the large fragmented PPP packets would be straight PPP packets lacking the multilink PPP header. Since they would be lacking sequence number information, they could get out of order at the remote end. So to use LFI, using more than one link in the MPPP bundle could lead to re-ordering issues.

This is where multiclass multilink PPP comes to the rescue. Cisco implements this using the ppp multilink multiclass configuration command. When turned on, the sequence number is maintained per class and so the interleaved small packets also would have the multilink header and so could be ordered correctly. This was defined in RFC 2686 and Cisco too has a good document that explains this.