Cisco Systems 1000 Series Manual De Usuario

here, you must also define a QoS policy with one or more priority classes, and attach the QoS to this 
interface using the service-policy output policy-map name command. This command classifies the 
priority traffic, that is interleaved by the MLP.

See the QoS and LFI configuration examples in the “Configuring Multilink PPP Connections” chapter 
in the 

When configuring MLP fragmentation on the various Cisco platforms, the functionality of MLP 
fragmentation and interleaving support on the various platforms may differ. This section explains the 
configuration options and their interpretation in the context of the Cisco ASR 1000 Series Aggregation 
Services Routers.

Based on the values of the MLP fragmentation configuration commands, the MLP feature calculates two 
values that are used during MLP fragmentation: link weight and maximum fragment size. These 
parameters are calculated for each member link in the bundle.

First, a link weight must be determined for each member link. The link weight indicates the number of 
bytes, and the MLP uses this value to balance the data amongst the links in the bundle. This parameter 
is especially important when the links in a bundle are of unequal bandwidth. The link weight is based on 
a combination of the bandwidth of the member link and the PPP multilink fragment delay value. If you 
do not configure the fragment delay value, a default delay value of 30 milliseconds is used:

Link Weight = (Member Link Interface Bandwidth in bps/8) * Fragment Delay

Configuring the fragment delay to a smaller value results in smaller fragment size because the fragment 
delay value determines the default fragment size on the member link. This, in turn, implies loss of 
bandwidth due to the added Layer 2 header overhead. This is important for broadband MLP, which can 
have Layer 2 headers of 4 to 58 bytes in length.

The default maximum fragment size must be calculated per member link. The default maximum 
fragment size used will be the lesser value obtained from either of the following calculations:

Link Weight – Multilink PPP + PPP Header Overhead (8)

Interface MTU – Multilink PPP Header Overhead (4)

After the default maximum fragment size is calculated, if you have configured the ppp multilink 
fragment size (maximum) command at the multilink, virtual template, or serial interface level, the 
default maximum fragment size is compared against the configured maximum value and is capped 
accordingly. If the fragment size is configured at the serial interface level and the multilink interface 
level, the serial interface configuration takes precedence.

Earlier, some Cisco platforms supported a legacy MLP fragmentation model that was enabled by default 
if all the following criteria were met:

Two or more active member links exist in the bundle.

All the member links have equal bandwidth.

No other form of multilink fragmentation or interleave commands are configured on the bundle or 
member-link interface.