New Canopy Deployment Questions

I am part of a group that will be in charge of deploying a canopy 900 network in the very near future. We will be using connectorized Canopy’s with MTI 11db Omni’s. Site to Site links will be by Redline. Management has decided to use PowerCode. (Hopefully they have their issues worked out by now. :hmm: ) We are in a relatively flat region of large corn fields interrupted by the occasional treeline/creek. Each site is around 8 - 10 miles apart.

The questions I have to worry about come from the deployment side of things. I’ll try to keep it as simple as possible:

1) What is the best scenario for color codes vs broadcast frequency?
Has anyone ran the same color code on all ap’s then set each pop within range to a different frequency, so they don’t overlap. You could then lock each customer down by frequency to the AP you desire. (I may be WAY off base here, but it seems to me it would be easier to remember one CC than 40)

2) IP Addressing:

Assuming a large complex network will be deployed with time, I would like to establish my IP addressing scheme with as much room for growth as possible. We are looking at an average of 40 to 50 clients per site. At what point does a routed network become a neccessity?
On Canopy networks, I have witnessed one provider leave default IP’s on all SM’s, and another provider assign a static IP to each SM. It seems like a management nightmare, and god forbid your ip schema need to change… Both providers run DHCP to the customer’s local device. How much of a role with Powercode play in the IP addressing?

This is a good start. I’ll be back for more later, I’m sure.

Thanks in advance!

I don’t recommend locking sms to aps via freq. Tf you have to change freqs you have to change every sm first. Leave all freqs checked and use unique color codes. if you have to change freq’s, the sms will follow the color code.

IP addressing depends on a flat network, or routed at each tower…

Does anyone have any experience with the PowerCode package?
Everyone is telling me to route the network from the start. I’m wondering how well powercode will work in a routed environment.

Jerry is right. Use color codes instead of frequency locking.

About routing or flat network, I think it also depends on how much you believe it will grow up.
I personally think that up to 500 SM flat network is much more simple to manage.

Anyway, I run a flat bridged network but every SM has NAT+DHCP enabled (in order to avoid any broadcast storm issue) exept from few bridged SM with L2 filters enabled (for the same reason).

In my network, however, I use a numbering scheme that should be quite easy to port to a routed network, in the future.
I use a class where the third number is the AP and the forth is the SM.
e.g. is an SM connected to the "AP 123"
it will be possible migrate to routed and preserve numbering, just by changing gw and subnet mask, and having one router for every tower.

+1 for color codes. Several times we’ve had to move freqs around and all we had to do was touch the APs - the SMs came back up to the AP they were initially set to. No muss, no fuss.

We have somewhat high-density towers (rural, but not middle of nowhere) with routed management and segregated subscriber access. On a management VLAN we have one subnet for switches and backhauls and several subnets for APs and CPEs that are for that area or deployment only (SkyPilot 5.8 and Motorola 900 gear on the same tower are different management subnets). We then have two subscriber VLANs because we are moving people from DHCP to PPPoE. We run all of our CPE in bridged mode, and on both types of connections the customer receives a public IP. In our area, this is what our customers expect both as a level of service and level of involvement from us - we just bridge it and route it, we don’t baby sit anything, and they have control over their connection.

Reasons for splitting things up like that are really management oriented - our SkyPilot network is smallish (~240 customers) and its EMS likes DHCP - so one /23 subnet there. Our Canopy stuff is taken care of by Prizm, for growth we assigned a /22 to each area (there are two). In these ranges so far we have only reserved the first 50 addresses for access points, the rest of the addresses are assigned by the EMSes for CPE.

We also have a legacy network leftover from when we purchased a local WISP. (We are actually an established telco with wired-ISP operations since sometime in 1995.) A lot of work has gone into improving this area which was using consumer-grade switching and performing centralized NAT for everybody. Basically it was a 1000+ quare kilometer open broadcast. It has been split into seperate VLANs per tower with their own management subnet and a hook into a PPPoE router since all of the Tranzeo gear is not VLAN-aware and doesn’t have any filtering capabiity in bridge mode aside from “disable communication between clients” on the APs.

But for you, it really depends on size. I strongly recommend seperating subscriber data from management data, once that’s done you can pretty much do anything you want with filters on the radios.

Not something you asked about, but how come your organization has decided to go with Omni antennas despite such a heavy load? Depending on your bandwidth offerings you will run out of available airtime quickly with just a single AP serving so many customers; I imagine noise won’t be big of an issue in the middle of cornworld.