Voice over Wireless LAN Solution Guide v1.0 December 2005
______________________________________________________________________________________________________
Page 21
As a new client on the network associates to an AP for the first time, it goes through the same
steps a wired client does, such as being put into a logical VLAN, issuing a DHCP broadcast,
receiving an offer, and then communicating on the network. In previous WLAN 2200 products, the
first parts were determined by local options on the WSS 2270 and SSID, meaning the WSS 2270
could only assign the user to a VLAN that was local to the WSS. This confined the IP address to
that local VLAN. Then, as the user roamed to other WSSs in other VLANs and subnets, its
presence was maintained back to the original WSS such that the IP address did not change. This
means that in a Distributed Campus architecture, the number of WLAN client subnets could be
many, and clients could potentially be in any subnet. Put differently, the main factor in assigning a
user to a subnet was simply determined by where he started out in the network, not by policy.
This not only made the network unnecessarily complex, but it made troubleshooting end-user
problems much more difficult. In such a situation, not only do you have to figure out where the
user is now, but you have to figure out which switch is currently the foreign switch and which is
his anchor switch. Another limitation to this approach is that WTM 2245s must be placed in every
subnet where phones might be used.
With the WSS 2300, initial VLAN and subnet assignments are determined by policy and that
choice is not restricted only to local VLANs and subnets. The previous example can be applied to
the WSS 2300 to illustrate the contrast. Now when a client associates to the network, the WSS
2300 can determine the proper VLAN/subnet assignment based on policy. If that VLAN is not
available locally, the WSS 2300 can find another WSS 2300 that has a connection to the
specified VLAN. The client is automatically tunneled back to that other WSS 2300, and their
DHCP discover packet is broadcast onto the remote VLAN. The “anchor” is determined by policy,
not by happenstance. As an example, all phones can be assigned to one VLAN/subnet and all
laptops can be assigned to another VLAN/subnet by policy. Even in a distributed campus, this
greatly simplifies the WLAN network. You can have as complicated an L1/L2 topology as you
want, while maintaining a very simple L3 design for client connectivity. This gives the flexibility to
integrate the WLAN 2300 series into any existing network, using either architectural philosophy,
while still allowing operational simplicity at Layer 3 and above.
The point of these examples is to provide some design recommendations based on this remote
VLAN assignment capability. Where possible, try to simplify the number of subnets that are used
for client devices. Even in a Distributed Campus architecture, you can have a few central subnets
for clients. As a general rule, Nortel recommends that you put IP phones, wired or wireless, in a
separate VLAN/subnet from data devices. This can be accomplished by providing one
VLAN/subnet for all WLAN telephony devices, as shown in Figure 7. The data client VLAN design
is an abstraction (though best practice is still to simplify). Maybe the WLAN data network has
many client subnets, or maybe one—that is unimportant in this context because the focus is
support of VoWLAN.