DSL backhaul provisioning

Mailman List Mirror (Read Only)
1

-----BEGIN PGP SIGNED MESSAGE-----

Getting to something more operational, we've been deploying DSL here
and there in onesies and twosies (well, even a dozen) alongside the
dialup POPs, and not bothering a lot about adding more capacity.

I've been keeping the number of modems to about 8 customers per modem,
and 200 modems per T1 (roughly 8 per 56K channel), just as a rule of
thumb. In our experience, this keeps the dialin lines busy at no more
than 20 minutes per day at peak, and the T1s average 70% utilization at
peak. Customers seem reasonably happy.

We just got a bid on DSL service saying they provision at 200 DSL per
T1. That seems awfully high to me, since DSL potentially runs a lot
more traffic than 56Kbps modems. I was planning at 24 per T1.

Who's right? We don't have enough aggregate experience with DSL, even
though we've been offering it for about a year now. Individually, the
chicken plant (our first installation), car parts factory, jeans
factory, bookstores, etc, don't give a very consistent usage pattern.

Any of the bigger folks have a rule of thumb?

2

We just got a bid on DSL service saying they provision at 200 DSL per

    > T1. That seems awfully high to me, since DSL potentially runs a lot
    > more traffic than 56Kbps modems. I was planning at 24 per T1.

This was hashed out on the ISP/C mailing list about a month ago. Most
providers are putting in between 200 and 250 DSL users per T1 of outbound
bandwidth. Think about it: $2K/200 is $10/user/month of bandwidth.
$2K/24 is $85/user/month of bandwidth. If you're selling your DSL lines
for $300-$400/month, it's fine to give people $85 of bandwidth, but if
your DSL costs the same as everybody else who's losing money on it, you
need to keep that cost down in the $8-$10 range.

                                -Bill

3

At 24 per T1, how are you going to make any money?