C-2
1/19/05
ATM Services Configuration Guide for CBX 3500, CBX 500, GX 550, and B-STDX 9000
Allocating Logical Port Bandwidth on CBX 500 Shared SP Threads
Shared SP Thread Example
Beta Draft Confidential
The 599.040 Mbps bandwidth value is available exclusively for user cell traffic.
Management and internal switch control traffic have the potential to use a maximum
of 11 Mbps of thread bandwidth, but this value is already factored into the total
available thread bandwidth. The total available thread bandwidth starts at 611 Mbps,
and once the NMS reserves 11 Mbps for management and control traffic, 599.040
Mbps remains exclusively for user cell traffic. At no time does management or
internal control traffic conflict with the 599.040 Mbps of user cell traffic. If user cell
traffic exceeds 599.040 Mbps, user traffic may be lost (depending on the QoS class of
the user cell traffic) if the following conditions exist:
•
User traffic is a lesser priority than the management and internal control traffic
•
User traffic exceeds the overall 611 Mbps thread capacity
This NMS enforcement of SP thread bandwidth only applies when the switch has two
IOMs installed on the same SP thread. If the switch only has one IOM on a thread, the
maximum possible logical port bandwidth for all ports on the IOM is supported by the
599.040 Mbps limit.
Shared SP Thread Example
When a switch has two IOMs installed on an SP thread, you will notice the NMS
enforcement of the SP thread bandwidth whenever you attempt to provision two
OC-3/STM1 cards on the same SP fabric thread. As you provision logical ports, the
NMS subtracts the assigned bandwidth from the 599.040 Mbps total. After you
provision four OC-3/STM1 logical ports on the first OC-3/STM1 card using the
maximum 149.76 Mbps of bandwidth, there will not be any bandwidth left for the
other OC-3/STM1 card and its logical ports.
Consequently, when you have two cards installed on the same fabric thread, Lucent
recommends that you allocate the bandwidth accordingly, across all of the IOM ports.
In this example, you would allocate approximately 75 Mbps to each of the eight
logical ports. This enables each logical port to support 75 Mbps of constant bit rate
(CBR) traffic, and consequently allows full use of the thread bandwidth.
Even when you use 75 Mbps per logical port, you can still oversubscribe the logical
port to overbook the VBR and UBR service classes on the port. For example, by
reserving 10% of each logical port’s bandwidth (that is, 75 Mbps) for UBR traffic and
overbooking the UBR bandwidth, hundreds of UBR circuits can be set up. Since UBR
circuits are not policed, these best-effort UBR circuits can potentially utilize the full
port bandwidth of each logical port, and consequently the full thread bandwidth.
However, at periods when the combined UBR traffic exceeds thread bandwidth, the
excess UBR traffic is dropped.
Summary of Contents for CBX 3500
Page 888: ......