Advertisement
Figure 4
Cisco 7600 Internet Router System Architecture
Distributed IP
Services via 1 or 2
PXF per Slot
The Cisco 7600 incorporates a centralized switching and distributed IP service application architecture. The
Supervisor 2 module with the policy feature card (PFC 2) and the multilayer switch feature card (MSFC2) provides
centralized packet switching. This switching complex can simultaneously apply security access control lists (ACLs),
apply QoS ACLs for traffic policing and marking, make policy decisions, and determine where to switch the packet,
all at line rate regardless of packet size and flow length. Distributed IP network services are implemented in
distributed OSMs, where each OSM has 32-Gbps bandwidth provided by a switch fabric module (SFM) and contains
up to two PXF IP service processors to enable IP services at 6 Mpps.
The Cisco 7609 has the option to use a 256-Gbps SFM or a 32-Gbps bus that provides centralized switching
performance of 30 or 15 Mpps, respectively. When the SFM is used, the Cisco 7609 can also support distributed
switching with performance up to 168 Mpps by using the Catalyst 6500 16-port Gigabit Ethernet LAN card, which
is powered by the Distributed Forwarding Card (DFC) and the SFM.
The Cisco 7606 has the option to use a 160-Gbps SFM or a 32-Gbps bus that provides centralized switching
performance of 30 or 15 Mpps, respectively. When the SFM is used, the Cisco 7606 can also support distributed
switching with performance up to 96 Mpps by using the Catalyst 6500 16-port Gigabit Ethernet LAN card, which
is powered by the DFC and the SFM.
The Cisco 7603 has only the 32-Gbps bus option and provides 15-Mpps centralized switching performance.
Distributed switching is not supported on the Cisco 7603.
Advanced PXF Technology
PXF is a Cisco patented "adaptive network processing" technology, which enables hardware-accelerated,
parallelized, programmable IP network services to be delivered in a short time to market. As shown in Figure 5, the
PXF IP services processor contains 16 networking-optimized CPUs, which are arranged in a 4 x 4 array. The four
columns can be visualized as four pipelines where data packets flow through from ingress interface to egress interface.
The four rows can be visualized as four IP service stages, where at each stage a different high-touch IP network service
All contents are Copyright © 1992–2002 Cisco Systems, Inc. All rights reserved. Important Notices and Privacy Statement.
Packet
Forwarding
MSFC2
SUP 2
PFC2
F
OSM
OSM
OSM
OSM
256 Gbps
Switch Fabric
Module
16 Gbps
Bandwidth
per Slot
Cisco Systems, Inc.
Page 5 of 8
Separate
Control and
Data Plane
Distributed
FIB Tables for
Services
6 Mpps IP
Services per
Slot
Advertisement
