|
Grex > Coop10 > #21: PLEDGE DRIVE: Buying Grex a Much Faster Internet Connection |  |
|
| Author |
Message |
| 25 new of 203 responses total. |
tsty
|
|
response 125 of 203:
|
Aug 13 05:18 UTC 1997 |
re #121. it didn't/doesn't have to, but hey, old news.
|
mcnally
|
|
response 126 of 203:
|
Aug 13 06:17 UTC 1997 |
re: putting money in an interest-earning account -- how would that
affect Grex's non-profit status and would the relatively miniscule
amount of interest earned be worth the staff time used filling out
whatever tax paperwork was needed?
|
valerie
|
|
response 127 of 203:
|
Aug 13 15:52 UTC 1997 |
This response has been erased.
|
drew
|
|
response 128 of 203:
|
Aug 13 19:12 UTC 1997 |
I saw an ad for a PCMCIA ISDN modem for around $400. On the general principal
that PCMCIA equipment tends to be more expensive than its desktop equivalent,
it should be possible to get the ISDN routers for much less than $650 apiece.
You might want to shop around.
|
steve
|
|
response 129 of 203:
|
Aug 13 21:05 UTC 1997 |
Good ones cost money. There is a *vast* difference between an at-home
use of a piece of equipment, and commercial quality continuously used
equipment. The ISDN routers we're looking at are the cheapest we dare
go with.
|
krj
|
|
response 130 of 203:
|
Aug 13 21:20 UTC 1997 |
An ISDN "modem" is not the same as an ISDN router.
|
toking
|
|
response 131 of 203:
|
Aug 13 22:31 UTC 1997 |
IWLTA <and no its not a lie> the checks in the mail
|
mdw
|
|
response 132 of 203:
|
Aug 13 22:39 UTC 1997 |
A real "ISDN modem" is pretty rare. It's a thing that will talk to
ordinary voice modems at the other end (via ISDN to the CO, then via an
ordinary analog voice line to the other end.) The major win is that it
gets rid of one analog hop. It's something an ISP might do, if ISDN
lines were cheaper than ordinary phone lines (they aren't) and if the
modems were cheaper (they aren't). You can also use ISDN for ordinary
voice conversations, and there are a few special features you can get
that way that you can't get with an analog line.
Electrically, ISDN looks like this:
central office
talks via U loop over 2-4 wires to
NT1
talks via T bus over 4-8 wires to
NT2
talks via S bus over 4-8 wires to
1 to 8 devices
In europe (and elsewhere) the phone company owns the NT1. In the US
(only) the subscriber provides the NT1. The T bus only supports 1
device, but the S bus supports up to 8 devices. The S & T busses are
otherwise similar; you only need an NT2 if you have more than one device
to hook up to the line. In the US, because the NT1 is something the
subscriber has to provide, it is common to find equipment that has this
built in.
ISDN comes in to flavors, BRI and PRI. PRI provides 23 B channels plus
1 D channel (or 30+1 over E1, found in europe..). BRI provides 2 B
channels plus 1 D channel. A B channel is a 64K channel, or one "voice"
connection. A D channel is a 16K channel, used for control purposes.
One common use for a PRI is for a PBX, or a private phone switch (such
as you might find in a small office). PRI's are also commonly used by
ISP's. BRI's are what you might find in a home. There are several
different standards for the control signals sent over the D channel;
hence some equipment that works fine in europe would be difficult to get
working here in the US (and visa-versa).
Because a BRI provides 2 B channels, for data purposes, it's common to
use both at the same time to talk to the same remote site. Various
incompatible protocols are available to do this, one is called
"bonding".
One ISDN data channel merely provides a 64K synchronous full-duplex bit
circuit. It does not provide for packet handshaking, compression, or
anything else. For internet access, it is common to run "PPP" over
ISDN. Some older ISDN internet devices did not provide for packet
compression, but newer ISDN devices commonly do so.
There are indeed a lot of ISDN devices available for the PC compatible
world. These devices commonly interface to the ISA, EISA, PCI, or
PCMCIA busses. Of course, these devices need software, in order to
function. Unfortunately, all of the vendors we've located for such
peripherals have fallen prey to market pressures, and *only* provide
software for Windows, Windows NT, Novell Netware and similar proprietary
environments. The problem with these environments is that they (a) cost
more, and (b) are less flexible, and (c) are not necessarily robust
enough to operate reliably (7/24) in a relatively hostile environment
(potentially hostile users, saturated network connection, unattended
operation.) The software for most of these cards is of course assuming
"personal" use, which means they're going to contain features like
part-time use, and colorful glossy brochures, but aren't necessarily
going to contain features like reliable automatic error recovery (a
pop-up window may be "good enough"), or subnet routing (for machines
connected via ethernet to the PC).
Another sort of ISDN device that is not hard to find is something that
connects an ethernet to an ISDN line. The Ascend pipeline is one such
product. These aren't particular flexible either, but they have
acquired a reputation for reliable unattended operation, and are
relatively simple devices unlikely to malfunction. For our purposes, we
want a device that includes the NT1 interface integrated, supports some
form of bonding and compression, and will support these features when
talking to whatever is at the other end. Right now, that would be a
clone of the same device, but in the future, it's desirable to pick
something that would be likely to work with an ISP too. Some older
pipelines would only route to a few devices on the local ethernet, and
not further, needless to say, that would not do at all for the
"internet" end of grex's internet connection. That would also not do if
we find in the future that we need to implement a firewall, or have more
than a few devices on the local ethernet.
There are lots of other ISDN devices out there. For instance, there are
ISDN devices that connect to an ordinary serial port (data rates may be
limited to the speed of the serial port), there are computers that talk
ISDN directly (some of the newer sun workstations come with an ISDN
interface), there are gobs of ISDN cards for all sorts of other busses
(including VME, like we have on the sun4 machines), and there are many
different routers that support ISDN (like what you'd find at an ISP).
Most of these devices are either a bunch more expensive, have limited
functionality, and/or would be a lot of work to get working on grex.
The next step up from ISDN is a T1 line. T1 is slightly cheaper than
PRI, and offers slightly more data througput (it runs over the same
physical lines as PRI). We located one vendor of T1 cards for the ISA
bus, that we (or I at least) really liked. They provided a freebsd
driver for their card, which would make it almost a no brainer for us to
get this going. It would be a really great deal *if* we could find
somebody willing to give grex T1 internet connectivity (if we provided
all the hardware), and we had about $3K startup + $250/mo monthly to
throw into the physical connection.
|
i
|
|
response 133 of 203:
|
Aug 13 22:44 UTC 1997 |
Re: #127
Ah. Sounds like the answer is that interest may not be paid on a commer-
cial checking account. Grex would have to maintain a seperate savings/C of
D/whatever account to earn interest. The standard "account types & quick
descriptions" sheet I have from my bank shows several non-checking accounts
for businesses that earn interest.
|
scg
|
|
response 134 of 203:
|
Aug 13 22:49 UTC 1997 |
Many ISPs are now using digital modems talking over PRIs or channelized T1
lines (kind of like a PRI, but without the signaling needed for ISDN data
connections) to talk to customers' analog modems, so that's pretty much the
ISDN modem concept Marcus was talking about in his first paragraph. That's
the only configuration the new "56K" modems will function in. One end of that
connection has to be digital.
|
toking
|
|
response 135 of 203:
|
Aug 14 00:41 UTC 1997 |
re 132: ouch..my head hurts
|
mcnally
|
|
response 136 of 203:
|
Aug 14 04:46 UTC 1997 |
I think a T1 connection would be a substantial overreach for Grex..
I don't think we have the hardware to support the users who would be
needed to support such a thing on an ongoing basis. The most common
cause of death for systems such as Grex is over-expansion, not under-
devlopment..
|
mdw
|
|
response 137 of 203:
|
Aug 14 07:03 UTC 1997 |
The hardware per se is not a problem. Even the current 4/260 would
benefit from a faster connection, and backtalk in particular would
benefit greatly. A faster network connection would mean that the
bottleneck would be in the CPU & disk - resources we can incrementally
improve - rather than in the network.
To put it another way, if we could get a T1 "for free", there's
absolutely no real reason why we shouldn't go for it. We have a perfect
technical example of this in operation - nether.net. The only remaining
question would be if the increased network bandwidth would change the
character of the system. Personally, I'm quite convinced it would not;
I've seen far too much evidence that the most important determining
factor of the "character" of the system is the people & traditions of
the people already on the system; I've seen this over and over again, on
grex, m-net, the well, chinet, and many other systems.
The real problem with T1 is the $ expense. There are two components to
this. The first are the capital and operational expenses of the link
itself. They are not cheap, but they *are* very compatible to the
expenses of 10 phone lines, and certainly would support far more than 10
users. Clearly, since we are already covering a very similar expense
for about 10 real phone lines, this is not out of line with grex's
current budget. The second is the expense of connecting the line to the
*internet* - *that* covers (essentially), the cost of a slice of an ISP,
their connection to any other ISP, and eventually, some portion of the
internet backbone. *That* cost, at commercial rates, is several times
the monthly cost of the raw T1 connection, and given grex's current
budget, would represent a considerable gamble, *at present* (to say the
least).
In the long run, I'm convinced that if grex remains viable, a T1 (or
equivalent) will happen. Hopefully, grex won't become potbound, but
will become larger, evolving technology such as the web will push the
need for better connectivity, and the costs of connectivity have been
and should continue to drop. When we first started grex, we knew that
"someday", if grex were to survive and remain viable, it would have to
get an internet connection; but there was no way we could afford it at
that point. Now, I think most of us could not conceive of grex *not*
having an internet connection. So, to me, the question is not "if", but
"when". There have already been a few points where it almost looked
like grex might get a deal that would realize T1-like connectivity, but
various things out of our control kept getting in the way. Hopefully,
someday, we'll find a deal that will realize better connectivity for
grex. In the meantime, I think we should keep in mind that ISDN is not
the ultimate in network connectivity, but just another step that we're
finally able to make.
|
mcnally
|
|
response 138 of 203:
|
Aug 14 17:42 UTC 1997 |
I won't argue against the claim that Grex will eventually need
(and be able to support) a T1, I'm just cautioning against the
"if you build it, they will come" school of system expansion.
Taking on large expenses in anticipation of a large influx of
users that never wind up coming has killed many a small public-
access system.
|
janc
|
|
response 139 of 203:
|
Aug 18 19:15 UTC 1997 |
Thanks to toking for sending in his $20 pledge for the ISDN connection.
Also, thanks to janc for belatedly handing in the other 20% of his pledge that
he previously forgot about. Some people can't be trusted with money.
|
signet
|
|
response 140 of 203:
|
Aug 19 12:53 UTC 1997 |
janc: are you speaking in third person in case he doesn't figure out the
connection?
|
dpc
|
|
response 141 of 203:
|
Aug 20 15:46 UTC 1997 |
What's the best guess on when we get the line?
|
valerie
|
|
response 142 of 203:
|
Aug 21 15:08 UTC 1997 |
This response has been erased.
|
dpc
|
|
response 143 of 203:
|
Aug 21 21:23 UTC 1997 |
Good show!
|
steve
|
|
response 144 of 203:
|
Aug 24 16:47 UTC 1997 |
I have a call into Ameritech on this, but I belive that both ends
have been installed. The Ameritech person who did the Pumpkin end
of the install was told the wrong building, but had my pager number
and called for help. That got straightened out quickly, thanks to
our wonderful Ameritech contact Ken Kiernan's putting my pager number
on all service work. ;-)
|
valerie
|
|
response 145 of 203:
|
Aug 24 23:41 UTC 1997 |
This response has been erased.
|
steve
|
|
response 146 of 203:
|
Aug 25 01:28 UTC 1997 |
I have an order in, but it won't be shipped out 'till Monday--actually
I was too late for shipment then, but it also let me look and make sure
that no one else ordered them. I see no evidence that anyone has.
|
valerie
|
|
response 147 of 203:
|
Aug 25 04:33 UTC 1997 |
This response has been erased.
|
scg
|
|
response 148 of 203:
|
Aug 25 04:42 UTC 1997 |
How are we doing on collected pledges so far?
|
awijaya
|
|
response 149 of 203:
|
Aug 25 14:02 UTC 1997 |
Hello Valerie, I hear about CC/Credit card problem. I will ask my
financial connection in NY for waiver/reduced cost. Is this againt
IRS rule/law? Perhaps the annual payment can be derived from
overseas interest. What about minimum amount for each month/year (turnover)
|