Multiclass Multilink PPP Enhancement (Cisco Systems)
Previous implementations of Cisco IOS Multilink PPP (MLP) include support for Link Fragmentation Interleaving (LFI). This feature allows the delivery of delay-sensitive packets, such as the packets of a Voice call, to be expedited by omitting the PPP Multilink Protocol header and sending the packets as raw PPP packets in between the fragments of larger data packets. This feature works well on bundles consisting of a single link. However, when the bundle contains multiple links there is no way to keep the interleaved packets in sequence with respect to each other.
The Multiclass Multilink PPP (MCMP) feature in Cisco IOS Release 12.2(13)T addresses the limitations of MLP LFI on bundles containing multiple links by introducing multiple data classes. Normal data traffic and delay-sensitive data traffic are divided into Class 0 and Class 1, respectively. Class 0 data traffic is subject to fragmentation just as regular multilink packets are. Class 1 data traffic can be interleaved but never fragmented. The next transmit sequence number, expected sequence number, unassigned fragment list, working packet, lost fragment timer, fast-switching mode, and all statistics are managed per class, rather than for the bundle as a whole.
Multiclass Multilink PPP Enhancement (pdf)
No comments:
Post a Comment