Cisco Systems 1000 Series Manual De Usuario

Although LFI is thought of as a single feature, it is actually two independent features within MLP. MLP 
link fragmentation allows larger packets to be Layer 2 fragmented by MLP, and the fragments to be 
distributed across the various member links in the MLP bundle. These fragments are MLP encapsulated 
and sequenced. These fragments are then collected, reordered, and reassembled at the peer termination 
point for the MLP bundle interface.

For more information about interleaving with QoS, see 

.

Interleaving enables you to reduce transmission delay on delay-sensitive voice, video, and interactive 
application data by interleaving it with the MLP fragments. When interleaving is configured, the packets 
on the bundle interface that QoS classifies as priority packets are interleaved. These priority packets are 
PPP encapsulated and interleaved with the MLP-encapsulated fragments or packets. When the peer 
router receives the PPP packets, they can be immediately forwarded, whereas, the received 
MLP-encapsulated packets have to be reordered and reassembled before being forwarded. While link 
fragmentation and interleaving can be configured on any multilink bundle, this LFI functionality is 
beneficial only on bundles of 1 Mbps or less. Packet transmission delays of higher bandwidth bundles 
are such that QoS prioritization of priority traffic should be sufficient to guarantee preferential treatment 
of the priority traffic without the need for LFI.

One downside of interleaving is that when there are two or more links in an MLP bundle, the order of 
the PPP-encapsulated packets cannot be guaranteed. In most applications sending data, such as, voice, 
video, and Telnet, this is not an issue because the gap between the packets on a given flow is large enough 
that the packets must not pass each other on the multiple links in the bundle. Since the order cannot be 
guaranteed for the priority PPP-encapsulated packets that are interleaved, IP Header Compression 
(IPHC) is skipped on any packet that is classified as priority-interleaved packet. IPHC continues to occur 
for nonpriority packets that are sent as MLP encapsulated because MLP guarantees reordering before 
the packets are forwarded to IPHC.

The Multi-Class Multilink Protocol (MCMP) (RFC-2686) addresses the issues related to ordering of 
priority-interleaved packets. Currently, the MCMP is not supported on the Cisco ASR 1000 Series 
Aggregation Services Routers.

MLP LFI must be configured on the Cisco ASR 1000 Series Aggregation Services Routers to enable LFI. 

In the context of interface multilink or interface virtual template, use any of the following commands to 
enable link fragmentation:

ppp multilink fragment delay (delay in milliseconds)

ppp multilink fragment size (maximum fragment size, in bytes)

For MLP using serial links, link fragmentation can also be enabled by configuring the ppp multilink 
fragment size (maximum fragment size, in bytes) command on the member-link serial interface. 

If the MLP bundle has only one active member link and interleaving is not enabled, MLP fragmentation 
is disabled. In addition, all the packets are sent PPP encapsulated instead of MLP encapsulated. When a 
second link in the bundle becomes active or interleaving is enabled, MLP and fragmentation is enabled.

If the ppp multilink interleave command is not configured, only MLP link fragmentation is enabled. 
To enable interleaving, you must also configure the ppp multilink interleave command at the interface 
multilink level or the interface virtual template level. In addition to configuring interleaving as indicated