Chapter 4 Switch Fabric; Switch Fabric Operation - Cisco CRS-1 System Description

Switch Fabric Operation

Ingress data packets are received at a physical interface on a PLIM and transferred to the associated MSC,
where the packets are segmented into cells for efficient switching by the switch fabric hardware. Each MSC
has multiple connections to each switch fabric plane, which it uses to distribute cells to each fabric plane.
On egress, cells are reassembled into data packets before being transmitted by the egress MSC.
The cell structure used in the Cisco CRS 4-slot line card chassis switch fabric is a Cisco design and is
Note
not related to Asynchronous Transfer Mode (ATM) cells.
Switch Fabric Operation
Several switch element components on each SFC perform the functions to implement each of the three
stages (S1, S2, and S3) of the switch fabric. Each stage performs a different function:
•
•
•
Speedup Function
A line card chassis can contain up to 4 MSCs, each with 140 Gbps of bandwidth. To provide 140 Gbps
of switching capacity for each MSC, the switch fabric must actually provide additional bandwidth to
accommodate cell overhead, buffering, and congestion-avoidance mechanisms.
Congestion can occur in the switch fabric if multiple input data cells are being switched to the same
destination egress MSC. Typically, little congestion exists between the S1 and S2 stages because there is
little or no contention for individual links between the switch components. However, as multiple cells are
switched from the S2 and S3 stages to the same egress MSC, cells might contend for the same output link.
To reduce the possibility of data cells being delayed during periods of congestion, the switch fabric uses
2 times (2x) speedup to reduce contention for S2 and S3 output links. The switch fabric achieves 2x
speedup by providing two output links for every input link at the S2 and S3 stages.
S2 and S3 Buffering
Buffering is also used at the S2 and S3 stages of the switch fabric to alleviate any additional congestion that
the switch fabric speedup does not accommodate. To ensure that this buffering does not cause cells to arrive
out of sequence, the MSC resequences the cells before reassembling them into packets. To limit the amount
of buffering required, a back-pressure mechanism is used for flow control (which slows the transmission of
data cells to a congested destination). Back-pressure messages are carried in fabric cell headers.
Failure Operation
The routing system can withstand the loss of a single plane of the switch fabric with no impact on the
system. The loss of multiple planes results in linear and graceful degradation of performance, but does
not cause the routing system to fail.
Cisco CRS Carrier Routing System 4-Slot Line Card Chassis System Description
4-2
Stage 1 (S1)—Distributes traffic to Stage 2 of the fabric plane. Stage 1 elements receive cells from
the ingress MSC and PLIM (or RP) and distribute the cells to Stage 2 (S2) of the fabric plane.
Stage 2 (S2)—Performs switching and the first stage of the multicast function. Stage 2 elements
receive cells from Stage 1 and route them toward the appropriate egress MSC and PLIM.
Stage 3 (S3)—Performs switching, provides 2 times (2x) speedup of cells, and performs a second
level of the multicast function. Stage 3 elements receive cells from Stage 2 and perform the switching
necessary to route each cell to the appropriate egress MSC and PLIM.
Chapter 4 Switch Fabric
OL-10805-10