On Mon, 09 Dec 96 07:17:43 -0800 Michelle Dick <artemis@rahul.net> said:
>I sent a message of typical largest digest length (40k) and it flooded
>the link at full-bandwidth (...) for 8 hours straight. Basically
>rendering my link useless for any other use. To upgrade to a 56k frame
>relay would run me, oh, maybe $500/month.
But you are mixing several things here. What you're really saying is that
your 28.8k dial-up line (which, given compression, probably gets you
about 56k of usable bandwidth for mailing list purposes) only costs you
$20/month plus another $20/month for the dedicated phone line, whereas a
56k flame delay line would cost you $500/month or 12.5 times more. I
couldn't agree more, and that's why I have a dial-up setup at home with
no plans to upgrade to flame delay, but that just tells us that the
$500/month option is not good value for money (at least in your case).
A better way to look at the situation is to say that you could spend
another $40/month and have a second dial-up line dedicated to the mailing
list function (let's set aside the issue of whether this can be made to
work on this or that operating system - we're talking bandwidth costs
here). For $40/month you would get roughly 56k of compressed bandwidth
and your mailing would go out in 8h. This is all list price - no special
deals. The charges are split roughly 50/50 between telco and access.
>The cost of this same usage at my ISP is about $6/month or $72/year.
Ok, so it actually costs your ISP $6/month to provide the service they
are charging you $20/month for. You are fortunate to be able to buy this
service at cost, and this is all fine and well, but under normal
circumstances the ISP would still want to make a profit. The figures I
used (other than being higher due to having estimated 300k/day) were
again based on list price, ie what you have to pay an ISP to get this
bandwidth. I had also included the telco charge, because you have to pay
for this as well. Finally, some ISPs have outrageous profit margins and
there is not all that much competition in this field. Let's just say that
some major ISPs have been SOOOO interested in getting a $100k contract
from us that in 3 weeks they didn't manage to issue a quote, and when
asked whether they have any interest in the sale, they just responded
that there are so many people placing orders that they can't always
answer everyone on time. Anyway, we have two issues here:
- At list price, the bandwidth costs you $40/month over dial-up (telco
included) and $15.55/month at the 10Mbps volume level (telco also
included). You can more than halve your costs by getting a fraction of
a bigger line, as expected.
- The actual cost to an ISP might be just $6/month, and they put the
difference between the access charge (ie telco excluded) and the $6 in
their pocket. That's life.
My point here is that the $6 figure is not realistic because there is no
reason for an ISP to want to sell at cost to a normal customer, and
certainly the $500:$6 ratio is not representative of a typical situation.
At list price you can save say $25/month = $300/year, which is not bad
but needs to be applied towards the purchase of cycles on a mail server
maintained by a third party that presumably also wants to make a profit.
To put it another way, if you have an existing PC with the necessary
horsepower to run the list and an existing dial-up connection that you
need anyway, and you can purchase the delivery service for $6/month,
there is no question that you have the best deal you will ever get. In a
traditional small/home business situation, however, the equation is
likely to be (in parentheses I'll show the costs with the PC amortized on
36 months):
1. Own MLM+MTA: $2500 PC + $40/month ($110/month).
2. Own MLM, remote MTA: $2500 PC + $15.5/month bandwidth + MTA charge
($85.5/month + MTA charge).
3. Remote MLM+MTA: depends on provider, probably $100-150/month.
Solutions #2 and #3 will provide better delivery times due to having more
available bandwidth, and are thus preferable in most cases. However I am
not convinced that #2 is cheaper than #1. #2 and especially #3 are likely
to be a bit more expensive due to the use of more expensive hardware.
While it may not be a big deal to this or that customer if his list is
down for 3-4 days while the local basement PC company fixes his machine,
the provider will have hundreds of lists on the box and needs to take
steps to prevent downtime. Whatever these steps are (more robust
hardware, backup machine, on-site spares, 24x7 support contract, etc),
they are going to cost money.
In my experience, most businesses do either #1 or #3. I do know of a few
sites that do #2, but there aren't many of them. ISPs don't seem very
interested in selling deliveries in large amounts, and there doesn't seem
to be much demand either. We've sold deliveries to a few customers at a
rate that I thought was very competitive, and they were happy but in the
end they brought it all in-house because they just wanted to have it
in-house for non-technical reasons. We don't even have a price list for
this service any longer because it is more work to keep the price list
updated than to issue custom quotes a few times a year.
Eric
Follow-Ups:
|
|
