User guide

New and Changed Information

Release Notes for Cisco 2500 Series for Cisco IOS Release 12.0 T

78-5563-07 Rev.H0

For Frame Relay, a new interface command, frag-pre-queuing, has been added that allows you to set

the queuing to be performed after the data segmentation. The command is available for Frame Relay

interfaces only. The syntax for this new command is the following:

frag-pre-queuing

no frag-pre-queuing

By default, this command is enabled, which allows only FCFS queuing at the interface level. If you

enter no frag-pre-queuing, you can configure weighted fair queuing, custom queuing, or priority

queuing at the interface level. Note that if you enter no frag-pre-queuing, you still must explicitly

configure the fancy queuing type on the interface.

For HDLC encapsulation, the queuing now takes place after segmentation when the voice-encap option

is entered. Weighted fair queuing, custom queuing, and priority queuing are now supported on an

interface configured for Cisco HDLC.

Flow-based WRED

This feature provides a mechanism to penalize the flows that do not respond to Weighted Random Early

Detection (WRED) drops. This feature is provided as an extension to the existing WRED functionality

and can be turned on after WRED is turned on.

Flow-WRED ensures that no single flow can hog all the buffer resources at the output interface queue.

With WRED alone, this can occur in the presence of traffic sources that do not back off during

congestion. Flow-WRED maintains minimal information about the buffer occupancy per flow.

Whenever a flow exceeds it's share of the output interface buffer resource the packets of the flow are

penalized by increasing the probability of their drop (by WRED).

Large Scale Dialout

Large scale dialout eliminates the need to configure dialer maps on every network access server (NAS)

for every destination. Instead, you create remote site profiles containing outgoing call attributes

(telephone number, service type, maximum number of links, and so on) on an authentication,

authorization, and accounting (AAA) server. The profile is downloaded by the NAS when packet traffic

requires a call to be placed to a remote site. Large scale dialout also takes advantage of features

previously only available for incoming calls, such as dialer and virtual profiles, Multichassis Multilink

PPP (MMP) support, and the ability to use an AAA server to store dial out attributes. MMP allows

NASes to be stacked together and appear as a single NAS chassis so that if one NAS fails, another NAS

in the stack can accept calls. Additionally, large scale dialout addresses congestion management by

seeking an uncongested, alternative NAS when the designated primary NAS experiences port

congestion.

Multilink Inverse Multiplexer

The Multilink Point to Point Protocol (MLP) Inverse Multiplexer feature allows you to combine

multiple T1/E1 lines in a Versatile Interface Processor (VIP) T1/E1 interface into a bundle that has the

combined bandwidth of the multiple T1/E1 lines. This is done by using a VIP MLP link. You choose

the number of bundles and the number of T1/E1 lines in each bundle. This allows you to increase the

bandwidth of you network links beyond that of a single T1/E1 line without having to purchase a T3 line.