I would like to connect mesh nodes to a large network of Cisco switches where it would be difficult to reconfigure the existing vlan assignments to use vlans 1 and 2. Is it possible to renumber the vlans used by the mesh node? For example:
vlan20 WAN
vlan100 LAN
vlan110 DtD
-mikeu
vlan20 WAN
vlan100 LAN
vlan110 DtD
-mikeu
In the case of the DTDLink vlan this is likely to be the case indefiently as its defined as part of the AREDN Protocol V3 Standard as being an 802.1q tagged vlan with ID 2.
The WAN Link is less stuck in stone but there isn't any supported method currently for changing it.
What to do will depend a bit on your environment.
DTDLink doesn't need to populate past the locally connected devices so it can generally be ignored for the corproate infastructure.
The LAN port is untagged and can be applied any vlan by the local switch infastructure. This leaves only the WAN vlan to be in question.
There are a couple of ways this can be done:
1) Since this is a Cisco shop check if the equipment supports "vlan mapping" (vlan translations.) If it does they can renumber the vlans as they come in/out of the switch to correspond with the local network infrastructure.
2) A "buffer" switch ban be placed locally, one of the low-cost netgear switches can be used to isolate the vlans locally for LAN+DTDLink and to feed out the WAN port as needed to the existing switching infrastructure.
There are also other methods (such as NAT) that rule out the need for using a "WAN" VLAN and the DTDLink not needing to populate past the directly connected devices can be ignored and blocked altogether at the existing switch gear and as such you only need to transfer the untagged ""LAN" packets which can be given any VLAN internally on the existing switchgear.
In addition there is potential for Q-in-Q vlan's if the environment needs to be spread out (note though DTDLink is officially not supported past directly connected devices so if were talking spanning a large campus fabric your moving closer to a newer feature being worked on for 'backhaul' networks which will be configured differently.)
Hope this helps you. As with most things network infrastructure wise there is often many ways to achieve the outcome and more specs may be required before a final solution is able to be given.
Currently I'm using a wrt54gs as a buffer to the Ciscos. The vlans 1 and 2 are already allocated for other traffic. There are separate ports that send untagged. So I do have a working solution but was hoping to integrate a bit more seamlessly.
It doesn't look like any of my Cisco switches support the IOS "switchport vlan mapping" command. The core 2970 switches and the NME-16ES-1G module in the router do support "switchport mode dot1q-tunnel" but the 2960, 2950, and 2940 edge switches do not. The mesh nodes are at the edge.
I'll give this some thought and post again if I come up with a better method.
-mikeu
I've never actually seen a device that does this either. Even looking at commercial AP's all the ones that I have seen for doing VLAN's over WIFI require a separate AP SSID for each VLAN (eg "802.1q over 802.11" doesn't exist)
Just found a slide deck from the 802.11 Working Group that mentioned that is exactly how they do it (SSID per VLAN) at IEEE conferences
Not sure if this is some technical reason off hand or not, but suffices to say we don't have support for it
If you need to pass some sort of traffic over the mesh that doesn't fit within the mesh infrastructure there is always the option to pass it through standard (unencrypted) tunneling (IP GRE Protocol is commonly supported) (this has also been a suggested method for linking two remote facilities together that need a logical LAN but need to transition the mesh for a backup link)
Mikeu,
You might try interfacing a netgear (with mesh node plugged into) with the cisco existing infrastructure to change vlan tags across them. This is Conrad's option #2 approach. For example:
on the netgear:
port 1: mesh node with vlans 1, 2, 100-LAN
port 2: vlan1, but untagged packets going both directions, to a port, e.g. port 2 on the cisco where the port tags them PVID=20. netgear PVID=1 for the reverse
port 3: vlan2, but untagged packets going both directions, to a port, e.g. port 3 on the cisco where the port tags them PVID=110. netgear PVID=2 for the reverse.
This means 2 cables between the netgear and the cisco router to map vlan 1 & 2 to the cisco vlan definitions. Vlan 110 tagged packets can be trunked over ether one of these cables. This approach is a hack with the cost of the extra netgear switch and 2nd cisco port in use to make up for the lack of ability to change vlan tags in the mesh node, but it should work for the extra ~$40. Cisco terminology is different, but PVID equivalent configuration.
Joe AE6XE