3com 5500-SI Manual Do Utilizador

A

PPENDIX

 D: 3C

OM

 XRN

n

All multihomed links and alternate paths must carry all VLANs, and packets 
must be tagged.

n

The Distributed Fabric is the STP root bridge.

n

Individual port members of each aggregated link must have VLAN membership 
manually configured before the aggregated link is set up. You must not rely on 
port members inheriting VLAN membership configuration from the aggregated 
link. (See “VLANs” on page 681 for more information.)

n

If you are using resilient links, these must be configured on the remote unit, 
not on the units within the Distributed Fabric.

If you follow the 3Com recommendations, should there be a unit or interconnect 
failure within the Distributed Fabric, traffic flow will be maintained at all times. If 
you want to know more detail about how the Distributed Fabric behaves in certain 
failure scenarios, see “How a Failure affects the Distributed Fabric” on page 684.

This section outlines the mechanisms that the Switch 5500 Family uses to 
determine the unit IDs for management purposes.

When a Fabric is created using the Switch 5500 the unit numbering can be 
determined in two ways.

n

You can manually assign unit IDs 1 to 8 to specific units using the 
change[self-unit, unit-id] to [1-8, auto-numbering] 
command from the System View. If you manually assign unit IDs to a Switch 
using the change command the IDs will be retained after a power cycle. 

If you add a unit to a Fabric that has previously been manually configured with 
a unit ID and this conflicts with an ID already within the Fabric, then the Switch 
with the lowest MAC address assumes the ID in question and the other unit 
will automatically renumber.

n

Fabric topology is ‘discovered’ and the units auto-number their IDs. 

Adding and removing units from the Fabric does not cause any renumbering to 
occur and the Fabric will continue to work normally. Renumbering only occurs 
when the Fabric is next power cycled if the units are configured to auto-number.

The unit LEDs will display the unit number in the Fabric, from 1 to 8.

The following example explains how to set up XRN in a network to gain maximum 
resilience using two Distributed Fabrics. The same process scales for larger 
networks if you are using multiple XRN Distributed Fabrics.

The example in Figure 180 shows a network with two Switches interconnected to 
create a single XRN Distributed Fabric in the Server Distribution section of the 
network, this Fabric in turn is connected to a Fabric within the Core of the 
network. The servers are multi-homed as are the Switch 5500 Fabrics to create a 
highly resilient network.