|
Grex > Coop8 > #134: Should Grex get a T1 connection? | |
|
| Author |
Message |
| 25 new of 146 responses total. |
krj
|
|
response 100 of 146:
| 
Nov 11 20:51 UTC 1996 |
T1 makes Backtalk fast enough to be useful; Backtalk opens Grex
conferencing to the hordes of web-people who have never run a CLI in
their life. So to me, the interesting (and unanswerable) question
is: how many future Backtalk users will become Grex supporters?
|
ajax
|
|
response 101 of 146:
|
Nov 11 22:12 UTC 1996 |
Backtalk is slow because of both a slow Internet link and a slow CPU.
Even with a T1, if Grex is particularly bogged down, Backtalk will be slow.
Every Grex user could benefit from the T1 (even people who don't telnet in,
but may use lynx or telnet out), so I don't think Backtalk users need to be
the primary source of T1 funding.
STeve, you mentioned we needed 40 more members to fund a T1. According
to my calculations, we need more like 30. Does someone have a breakdown of
how 40 was estimated? Also, since Grex has a bit of surplus, we may already
have a few of the necessary 30 "spare" members, though there are certainly
no shortage of other things on which we've been spending surplus cash. :-)
|
popcorn
|
|
response 102 of 146:
|
Nov 11 23:01 UTC 1996 |
Given a slow link, I think Backtalk is actually a better way to read the
conferences than Picospan is. Aside from getting to skip the telnet queue,
Backtalk has the advantages that 1) when you press the Page Down button, the
next page is already there. You don't have to wade through netlag to see it.
And 2) when you compose a response, you're doing it off-line. Editing is
a lot easier when there's no netlag.
|
mdw
|
|
response 103 of 146:
|
Nov 11 23:15 UTC 1996 |
(This isn't a good place to talk about whether or how m-net might
implement such a limit, so let me just say that if you want to do it,
the logical place to do it is the "login" program - to which you have
source. Login will need to check *after* it asks for and checks the
password, and will need to go into some sort of "waiting" state if there
are already too many people of the appropriate sort on. One good way to
do the "counting" is with a SVID semaphore - this is how AIX (for
instance) counts users. You will also need to allocate extra pty's for
this, so you should expect to end up with (say) 32 guest ptys, 32 patron
ptys, & at least 64 "waiting" pty's, for 128 total. You should also
talk with your OS vendor about what it would take to support more than
256 PTY's on your machine, and now would be a good time to start
worrying about long uid support.)
My unscientific estimate is that we should be hoping to attract about a
hundred extra members, for around 200 total, with the T1. Besides the
T1 proper, we'll also need to think about funding other improvements.
|
popcorn
|
|
response 104 of 146:
|
Nov 11 23:27 UTC 1996 |
Here's part of the treasurer's report from the September 25, 1996 board
meeting:
> 2. Treasurer's Report
>
> For planning purposes, Mark gave a presentation using a flip chart, which
> showed our income versus expenses over a 36 month period. This is important
> because our income fluctuates greatly on a monthly basis, but our finances
> are very stable over a period of several years. Here is the gist of it --
>
> income
> $500/month constant income from memberships
> $115/month JCC sakes (averaging 9 sales over 36 months)
> $ 60/month auction (assuming one auction per year)
> $ 10/month T-shirts and mugs
> -------
> $685/month total income
>
> expenses
> $275/month phone (excluding Internet link)
> $ 10/month miscellaneous
> -------
> $285
>
> This leaves $400/month for rent, i-net connection, electricity and and
> hardware costs. Historically, over 36 months, we have spent an average of
> $185/month on hardware maintenance and upgrades. This excludes money spent
> from earmarked funds, such as the disk fund-raiser.
>
> If we allow $185/month to continue to be allocated for hardware costs, we
> are left with $215/month to spend on i-net, electric and rent. The purpose of
> this presentation was for planning and deciding how much we could afford to
> spend on these items as we pick our next home.
>
> Steve Weiss pointed out that we might choose to allocate a smaller amount
> for planned hardware upgrades and maintenance.
Since we now spend $83 on electricity, and $60 on rent, that leaves $215
minus $143, or $72 per month for the Internet connection. Assuming $250
is the monthly telco cost of an Internet connection, if we get the 60 month
contract that STeve talked about, this means we need an additional $250 minus
$72, or $178 per month for this T1 offer. Each membership brings in $5 or
$6 a month, depending on whether it is a monthly or an annual membership.
I'm going to figure $5 per member per month, because it costs some money to
deposit each monthly check from a monthly member.
$178 per member divided by $5 per member per month is 36 new members.
(Disclaimer: I did all the math for this item in my head. Use a machine to
recheck it if you want to be sure of it.)
|
steve
|
|
response 105 of 146:
|
Nov 12 02:14 UTC 1996 |
Valerie's math sounds better than mine, Rob. So lets use the
36 member figure.
|
ajax
|
|
response 106 of 146:
|
Nov 12 05:42 UTC 1996 |
I see. I was estimating the cost of a T1; #104 estimates the cost of
a T1 above our current rate of income, based on average past income/
expenses. Our major departure is using different assumptions for the
cost of T1 service. I included installation fees, equipment costs, and
the $2400 donation during the first year, which Mark did not. In response
#65, I came up with a net one year cost at around $2200, or net three
year cost at $6100 (or $6800 if we need some equipment). #104 assumed
$250/mo Internet costs, or $3000 per year, which is a fair amount higher.
(Among other things, it doesn't factor in that we'd cut $40/mo by
canceling our current Internet service).
Combining my cost-of-service assumptions of $170-$190/mo, with Mark's
current "spare cash" estimate of $72/mo, I think we need an additional
$100-$120/mo, or 20-24 members, that we wouldn't ordinarily get.
As Marcus pointed out, if we want to expand with other improvements,
we'd need more money/members, and as srw pointed out, if we don't have
enough money/members, we could cut back on some expenses normally spent
on hardware improvements.
|
rcurl
|
|
response 107 of 146:
|
Nov 12 05:53 UTC 1996 |
I'd suggest building some safety factor into this "bridge to the future".
Maybe, a member base of 150? What does anyone think doing this will do to
the member base - increase it or decrease it? Everything that has been
done so far has left the membership essentially unchanged - maybe that's
what would happen, unless some fundamental change were made to make Grex
membership more attractive.
|
nephi
|
|
response 108 of 146:
|
Nov 12 07:28 UTC 1996 |
(Catching up . . . )
Info points that folks may find interesting in making their decision:
In mail to baff, one of the folks wanting to donate for the T1 said
that s/he was willing to pay $1200 up front in one lump-sum. I have
a suspicion that the other may do the same if it proves necessary.
Additionally, one of the donors has said that s/he would be willing to
financially support the T1 beyond the first year, although it couldn't
in good faith be guaranteed.
|
popcorn
|
|
response 109 of 146:
|
Nov 12 16:31 UTC 1996 |
Re 106: Actually, #104 did take into account the $40/month that would be saved
on ICnet phone lines: Mark said $x would be left over, after other expenses,
for rent, electricity, and the Internet connection. I didn't add in the cost
of the ICnet phone lines in Grex's monthly costs; neither did Mark.
You've probably said it elsewhere, but where did the $170-$190/month cost of
service figure come from? I used the monthly Ameritech charge for a T1
connection, given a long-term contract, from the numbers STeve got from
Ameritech. STeve's numbers are substantially higher, even for a long-term
contract.
Re 107: Actually, some improvements *have* brought in new members, if I
remember right. These were increases in the number of modems.
To the anonymous donors: Thanks!
|
rcurl
|
|
response 110 of 146:
|
Nov 12 20:32 UTC 1996 |
It would be useful to see a tabulation of number of member and "mileposts",
both as a function of time. The latter would be things like Valerie mentions
- "modems increased from x to y", etc. Membership has certainly not tracked
*use*.
|
ajax
|
|
response 111 of 146:
|
Nov 12 22:15 UTC 1996 |
Re 109, my numbers were calculated in #65. As I said there and in 106,
they include a number of factors besides Ameritech costs, including
equipment, the anonymous donors, installation, and money saved saved on
current Ameritech costs. Since the $40 savings was built into Mark's
figure, that brings my estimate back to about 30 members. Personally,
My main disagreement with the other type of estimate is that using
using "average past donations," while ignoring the anonymous $2400
donation, exaggerates the cost.
You can add whatever numbers you want for a safety net and further
expansion; Rane suggests 150 members, Marcus suggests 200 members. Good
goals, but I think for the purposes of this item, it's most useful to
estimate the money/members required for a T1, which is closer to 130.
Rane, I've been curious about the membership effect of upgrades,
particularly after the CPU earlier this year. I'll dig through the
old financial reports and post a summary of them.
|
janc
|
|
response 112 of 146:
|
Nov 12 23:32 UTC 1996 |
(At some point I posted a message in the motd saying we needed 40 more
members. That was a number I made up as "more than enough to do some
kind of upgrade of the net connection. It wasn't the result of any
fancy calculation and didn't mean much.)
|
srw
|
|
response 113 of 146:
|
Nov 15 06:10 UTC 1996 |
I agree with Rob (ajax)'s comment that using average past donations and
ignoring the $2400 donation exaggerates the cost. At 130 members I believe
it will be sustainable. Actually it may be at a lower number, because we
can also make up some of the extra with fundraisers, but I don't like to rely
on such activity. I'm quite confident that we can achieve the membership level
we need to support this if we can get over the difficulty of startup.
I am not the anonymous donor. I haven't offered anything for the T1 yet,
although I had planned to. I am disturbed that some might consider that my
support for the T1 has some mercenary component. It does not.
Of course I want Backtalk to look good, but I want Grex to have the T1 because
I think it will be good for Grex, not for Backtalk. Backtalk is already
installed on HVCN, connected to the net by a T1. This is not new.
|
tsty
|
|
response 114 of 146:
|
Nov 15 07:00 UTC 1996 |
backtalk *looks* good anyway. it just bogs the TinyLink more.
|
chelsea
|
|
response 115 of 146:
|
Nov 15 09:47 UTC 1996 |
No mercenary component was suggested, Steve. What was being
suggested is that any decision to go ahead with a T1 line
be made as objectively as possible simply based on Grex's
needs. Stessing this is especially important since the
members are not privy to all information regarding this
deal.
|
ajax
|
|
response 116 of 146:
|
Nov 15 19:56 UTC 1996 |
Has it been determined yet what equipment Grex would need to purchase
to enact the T1 proposal? Maybe the ISP would provide equipment, but if
not, it could cost a fair amount of money, conceivably more than $1000.
I think it would be really useful to know this before next Wednesday's
board meeting.
Mary suggested earlier that a membership vote should be held concerning
the T1. Given the time frame to take advantage of Ameritech's installation
discount (the line has to be installed, or maybe ordered, by 12/31), I
think the board should make the decision. A member could still request
a vote to reverse whatever the board decides, and if it's concluded by
12/31, Grex could probably cancel the line installation without cost, or
order its installation without cost.
|
aaron
|
|
response 117 of 146:
|
Nov 16 16:44 UTC 1996 |
I should point out that there is some of the same thinking here that is
problematic on M-Net -- i.e., there isn't much thought given to new sources
of income. Now, it may not be easy to come up with great new sources of
income, but it is certainly wise to do so. Notably, without 501(c)(3), Grex
has significantly more freedom in that regard than M-Net. (Welcome to the
Grex internet shopping mall. ;)
|
scg
|
|
response 118 of 146:
|
Nov 16 19:37 UTC 1996 |
Jan has proposed several times that we should have various Grex merchandise,
like shirts, coffee mugs, tote bags, etc. available for sale on our web site
all the time. That sounds like an excellent idea to me. I'd also be very
interested in hearing any suggestions anybody else has for fund raising
programs. I'm afraid fund raising isn't something I've ever been very good
at, so I'm not sure how much I'll be able to help with it, but it would be
great if somebody could pick that up as a project and work on it.
|
nephi
|
|
response 119 of 146:
|
Nov 17 00:19 UTC 1996 |
If someone could send me pictures of any Grex murchandise that
is currently available, I'd be willing to create a web site for
selling the stuff.
I think it is a terrific idea, and one that could bring in a
good amount of money.
|
dang
|
|
response 120 of 146:
|
Nov 17 23:14 UTC 1996 |
The point has been raised, and I agree with it, that if we start to
permenantly sell stuff, we should have a sales tax license.
|
kerouac
|
|
response 121 of 146:
|
Nov 19 00:24 UTC 1996 |
I think Backtalk works great on HVCN mostly but is too slow on Grex for the
most part to be practical at this point. Right now I only use Backtalk on Grex
when I want to post. If I have 15 or 20 items to read through, I do NOT want
to use Backtalk as it presently could take eons. If Grex can run faster so
more people can use Backtalk, Grex will end up eventually with more members.
(As a side drift, Grex is currently down for mail processing, but I'm posting
through Backtalk, which runs much faster when nobody is on. I think when Grex
is down for mail purposes, the note at login should direct users to go to
Grex's webpage and Backtalk if all they want to do at current time is
conference)
|
dpc
|
|
response 122 of 146:
|
Nov 19 02:50 UTC 1996 |
A post through Backtalk runs up against the left and right edges of
my screen, and words are split between lines, FWIW.
|
scg
|
|
response 123 of 146:
|
Nov 19 05:50 UTC 1996 |
Uh, Richard, did it occur to you that there might have been a reason I stopped
letting people log in while trying to get Grex caught up on mail processing?
The goal there was to take as much non mail load off the system as possible,
so that mail could get caught up quickly and the system could be opened back
up. If I had wanted to do something about backtalk, it would have made a lot
more sense to turn off httpd, thus turning off backtalk, than it would to tell
people to use it.
|
ajax
|
|
response 124 of 146:
|
Nov 19 07:33 UTC 1996 |
(Back to T1s....) I'm curious if a fractional T1 (or FT1) is an option.
Most ISPs offer such service. A full T1 combines 24 DS0 (64 Kbps) channels
into one DS1 (1,544 Kbps) line (24 * 64 Kbps + 8 Kbps is used for timing).
Fractional T1s use the same physical lines as a T1, but provide two or more
DS0 channels. Ameritech offers what they call "384 Service," which is
six DS0 channels, or 384 Kbps, over the same type of dedicated line as is
used for a full T1. This would presumably cost less per month, though I'm
not sure by how much (I expect it's still the majority of the price of a
full T1, but even a 20% discount might make it attractive). To handle
fractional T1s, I think you need a CSU/DSU on each end of the line (for a
full T1, you can sometimes use just a CSU, and we still don't know what
Grex would need on one or both ends of the proposed T1 connection).
I checked some web sites of ISPs that provide T1 links in the 313 area
code. MSEN offers 128k, 256k, and 384k FT1s, while CICNet just lists
"shared DS1" in their "The List" listing. In addition to whether the ISP
in question would offer service for a fractional T1, another issue is
whether the anonymous donors would contribute as much (or anything) toward
a fractional T1.
Concerning startup equipment, MSEN normally charges $1800 for hardware,
while CICNet charges $7500 for equipment plus service startup (that's for
a turnkey solution, with a Cisco router, CSU/DSU, and network management
equipment which allows them to monitor line conditions remotely).
(Prices are from their web pages.)
To me, the initial equipment costs are the deciding factor on whether
the tentative T1 proposal outlined in #0 should be accepted.
|