High Availability - Cisco CRS-1 System Description

Cisco CRS Carrier Routing System 8-Slot Line Card Chassis Enhanced Router Overview

High Availability

The CRS routing system hardware detects, isolates, and recovers from a broad range of faults, and provides
failover mechanisms to redundant hardware. The control plane is a central element in achieving high availability,
as it must isolate failures and direct failover events, both in the data plane and in the control plane.
To ease serviceability, the chassis identification displays and the critical, major, and minor alarm indicators
are clearly visible. Each line card, RP, and switch fabric card has an alphanumeric display and green OK LED
to show current board status. Environmental conditions, including temperature and voltage levels, are monitored
by several internal measurement points and reported to the routing system operator.
The RP cards function as the system controller in a Cisco CRS 8-Slot Line Card Chassis. Note that the PLIMs
are connected to the control plane through their respective line cards.
The control plane provides switched point-to-point Fast Ethernet (FE) connections between the control
processors on various routing system components. This allows for control plane network messages, and some
other paths for system communication. The dual RP cards and midplane FE traces provide redundant
connections between all cards in the line card chassis. Most cards or modules contain a service processor (SP)
module that provides the communication for that device within the control plane.
Some of the important functions and implementations of the control plane are listed in the following sections.
Online Insertion and Removal (OIR) Detection
Every line card, switch fabric card, power module, and so on provides a presence-detection signal to the
system controller function on the RP cards. This dedicated hardware signal indicates the physical presence
of a card in every slot. The presence-detection signal allows the Cisco IOS XR configuration software to
quickly detect online insertion and removal (OIR) events, and identify cards that have been inserted but cannot
communicate over the control plane.
PLIM Inventory
Every PLIM slot is probed by the master RP to get the board ID and type and other inventory information.
The RP can read an identification chip on each PLIM, even if the PLIM is not powered on. The PLIM inventory
chip can be accessed by the RP, whether or not a line card is plugged into the MSC slot associated with the
PLIM.
RP Active and Standby Arbitration
Both RP slots are directly connected by dedicated midplane signals to special hardware arbitration logic.
During the boot process, this logic selects one of the RP cards to be the master (active) device; the other RP
functions in standby mode. See
The control plane verifies that the arbitration logic selected only one RP as active (due to an unusual
circumstance in which both RP cards might mistakenly be made active).
After hardware arbitration, software should verify single RP mastership through the control plane FE messaging.
The arbitration hardware could elect two masters due to an unusual hardware fault. The control plane FE
provides a redundant path so mastership can be verified absolutely.
Node Reset
Each RP has a dedicated reset line to every node in the chassis. Nodes include MSCs, RP cards, and fabric
cards. The reset lines fan out from each RP and are connected to the SP on the node cards. Only the master
RP can assert these reset lines; the standby RP reset lines are isolated by the RP arbitration logic. The reset
lines allow the RP to force a board reset from hardware, and is used only if a board does not respond to control
Active and Standby Arbitration, page 6-4
Cisco CRS Carrier Routing System 8-Slot Line Card Chassis Enhanced Router System Description
High Availability
for more information.
9