Cisco Systems MGX-FRSM-HS2 Manuale Utente
Frame Service Module Features
Cisco MGX 8850 Installation and Configuration, Release 1.1.00, Part Number 78-6186-02
6-20
Four-Port Unchannelized Frame Service Module for V.35
The MGX-FRSM-HS1/B provides unchannelized Frame Relay service across four V.35 lines. The
maximum throughput for the card is 16 Mbps. The maximum rate on a line is 8 Mbps. Without the
cost of a T3 or E3 card, the MGX-FRSM-HS1/B provides greater that T1 or E1 speeds on a port as
well as a choice of 50 line rates in the range 48 Kbps–8 Mbps.
maximum throughput for the card is 16 Mbps. The maximum rate on a line is 8 Mbps. Without the
cost of a T3 or E3 card, the MGX-FRSM-HS1/B provides greater that T1 or E1 speeds on a port as
well as a choice of 50 line rates in the range 48 Kbps–8 Mbps.
Frame Service Module Features
This section first lists the features common to all FRSM models then lists the features of each model.
All FRSMs support:
All FRSMs support:
•
Frame Relay-to-ATM Network Interworking (NIW) as defined in FRF.5.
•
Frame Relay-to-ATM Service Interworking (SIW) with or without translation as in FRF.8.
•
Frame forwarding.
•
ATM Frame-UNI.
•
Maximum frame sizes of 4510 bytes for Frame Relay and 4096 bytes for ATM-FUNI.
•
Per-virtual-circuit (VC) queuing in the ingress direction (towards the Cellbus). Traffic arriving at
the network on a connection has a dynamically assigned buffer at the entrance to the switch.
Buffer size depends on the amount of traffic and the service-level agreement (SLA).
the network on a connection has a dynamically assigned buffer at the entrance to the switch.
Buffer size depends on the amount of traffic and the service-level agreement (SLA).
•
Advanced buffer management. When a frame arrives, the depth of the queue for the LCN is
compared against the peak queue depth scaled down by a specified factor. The scale-down factor
depends on the amount of congestion in the free buffer pool. As the free buffer pool begins to
empty, the scale-down factor is increased, preventing an excessive number of buffers from being
held up by any single LCN.
compared against the peak queue depth scaled down by a specified factor. The scale-down factor
depends on the amount of congestion in the free buffer pool. As the free buffer pool begins to
empty, the scale-down factor is increased, preventing an excessive number of buffers from being
held up by any single LCN.
•
Multiple, priority-level queuing to support class of service on the egress. The FRSM services
egress queues according to a weighted priority. The priority depends on the percentage of logical
port bandwidth needed by all connections of a particular type on a port. The FRSM supports a:
egress queues according to a weighted priority. The priority depends on the percentage of logical
port bandwidth needed by all connections of a particular type on a port. The FRSM supports a:
— High-priority queue
— Real-time Variable Bit Rate (rt-VBR) queue
— Common queue for non-real-time Variable Bit Rate (nrt-VBR) and ABR connections
— UBR queue
•
Initial burst per channel. After a period of silence, the FRSM sends a configurable number of
bytes at a peak service rate.
bytes at a peak service rate.
•
The ForeSight option. This Cisco mechanism for managing congestion and optimizing
bandwidth continuously monitors the utilization of ATM trunks. It proactively adjusts the
bandwidth for connections to avoid queuing delays and cell discards.
bandwidth continuously monitors the utilization of ATM trunks. It proactively adjusts the
bandwidth for connections to avoid queuing delays and cell discards.
•
Consolidated Link Layer Management (CLLM), an out-of-band mechanism to transport
congestion related information to the far end.
congestion related information to the far end.
•
Dual leaky bucket policing. Within the basic parameters such as committed burst, excess burst,
and CIR, incoming frames go into two buckets: those to be checked for compliance with the
committed burst rate and those to be checked for compliance with the excess burst rate. Frames
that overflow the first bucket go into the second bucket. The buckets “leak” by a certain amount
to allow for policing without disruption or delay of service.
and CIR, incoming frames go into two buckets: those to be checked for compliance with the
committed burst rate and those to be checked for compliance with the excess burst rate. Frames
that overflow the first bucket go into the second bucket. The buckets “leak” by a certain amount
to allow for policing without disruption or delay of service.