I am looking at setting up a Canopy network, but am finding the scarcity of realistic sample logical network diagrams to be a bit frustrating. (One AP w/ an IP and three SMs ported into a router? Come on…) Can anyone provide an example of configurations that do not work, or that you found created problems.
I also stumbled across the recommendation that the logical addressing reflect the flow of data across the Canopy network for best performance. Since Canopy essentially functions as a layer 2 pipe, why should that matter?
As an example the online docs state that there should be several subnets, but they group elements based on the type of hardware, ie a subnet for BHs, one for APs of different freqs, etc.
To me that doesn’t make much logical sense. For example
x.x.1.1-254 being just a lumped group of BHs, or
x.x.2.1-254 being a list of 2.4GHz APs (where x.x is some private IP)
It makes more sense in my mind to link the elements based on data flow/location.
For example, with an AP cluster fed by a pair of BHs
x.x.0.1-254 subnet for servers: BAM, NMS, etc
x.x.10.1 BHS <-> x.x.10.2 BHM
CMM @ BHM x.x.19.1
APs @ BHM x.x.11.1 => AP1
x.x.12.1 => AP2
x.x.13.1 => AP3
x.x.14.1 => AP4
x.x.15.1 => AP5
x.x.16.1 => AP6
with SMs attached to the appropriate AP. for example an SM attached to AP1 would have an address like x.x.11.2 (2-254) being possibilities.
Would there be any problems with this layout? Expansions could be handled by incrementing the third octet. For a second tower the same layout, but with x.x.20.1/2 being the BHs, etc.
User address space would be a Class C network, with the appropriate gateway configured on a router.
I guess what I am looking for is a solution that is going to provide relative simplicity in maintenence, but still provide some kind of logical grouping to the elements of the Canopy network to aid in identifying which modules are physically located together, or function as part of the same basic link.
We use this type of addressing.
Tower 1
172.18.1.1 Router
172.18.1.2 cmm
172.18.1.3 AP 1
172.18.1.4 AP 2
172.18.1.5 AP 3
172.18.1.6 AP 4
172.18.1.7 BH
172.18.1.8 BH
Tower 2
172.18.5.1 Router
172.18.5.2 cmm
172.18.5.3 AP 1
172.18.5.4 AP 2
172.18.5.5 AP 3
172.18.5.6 AP 4
172.18.5.7 BH
172.18.5.8 BH
And so on.
This allows us to seperate all towers and still keep a logical on where the device is located.
I’m assuming you are using NAT then?
We are trying to keep nat on at all times. If the customer needs it off we use dmz on the antenna.
I may have missed it, but there is no option to do port forwarding on the SM is there? All or nothing… heh
Are there any examples w/o NAT? How much address space is typically left for SMs? The AP can 100max w/ HW scheduling, 200 w/o, but even 100 is probably an unrealistically high number of customers on any one AP.
What is the max number of subscribers that anyone has connected to a single AP 20Mbps ver, or 900MHz, or both?
Neteru
My addressing scheme is very much like you have described.
APs 10.10.x.1 each AP increments x. (Radio IP LAN1)
I start the SMs at 10.10.x.10 leaving the 2-9 for aux equipment. (RF Public)
I try to use DHCP server and NAT whenever I can.
NAT Public = Static IP
NAT Private 10.33.1.1
DHCP start address 10.
If you would like a copy of my addressing diagram PM me with your email address and I will prepare a PDF for you.