response 26 of 95:

Sep 24 17:03 UTC 2003

I think that was seven hours of busy signal over *three* months.  
The consensus was we'll drop to four now and then
evaluate how it's going for those dialing in.  We can drop
additional lines, without charge, but there is a $42 fee
to add a line.  I think I've got that right.

response 28 of 95:

Sep 24 23:03 UTC 2003

I hope this isn't the beginning of the end for dialup. I don't imagine I'll
ever use dialup unless i actually move to Michigan (unlikely, since I'm
presently firmly rooted in the UK, or at least the EU), but they're handy for
people who live in Michigan and want to connect when there's a wait for an
Internet line.

I do support the idea of reducing the number of lines, though, since some seem
never to be used; not much point in disagreeing, now that it seems to be a
fait accompli :-).

response 30 of 95:

Sep 25 12:52 UTC 2003

They're more than handy for people who want to avoid the telnet queue.  There
are a number of people who have no other internet access.

response 32 of 95:

Sep 26 00:53 UTC 2003

I know.  I mentioned it.

response 34 of 95:

Sep 26 04:21 UTC 2003

I've been put it in occasionally, over the past month.  I don't know why,
though:  when I get on, the actual usage is pretty reasonable.  Right now,
we have "0 waiting, 28 remote + 2 local users; 72 max remote users".

response 36 of 95:

Oct 4 02:43 UTC 2003

That's assuming that telnet users and dial-in users see the same
throughput and delay and present equal loads.

response 38 of 95:

Oct 4 12:50 UTC 2003

Re 37:  No.  You need to spend some time on the vandal-smacking patrol to
understand where the bandwidth goes.

response 40 of 95:

Oct 5 05:50 UTC 2003

You misunderstood me.  Actually, the fault is mine; I was unclear.  It
could be argued that dialup users create more system overhead than if
they connected via the network (the cost of handling serial interupts
is high).

response 42 of 95:

Oct 5 13:26 UTC 2003

The "serial interrupts" argument might have been true back years ago when we
were still using the multiport serial card.  I vaguely recall that Marcus and
STeve would occasionally argue that point.

The current situation is that modem users connect to our terminal server,
which then gives them a telnet connection to Grex.  Regardless of how the
modems connect, it's never been the modem users who download
eggdrop/bitchx/etc, using up network, disk, CPU, and staff resources.

response 44 of 95:

Oct 6 14:26 UTC 2003

OK, I called Ameritech and told them to implement the motions the board
passed at the last meeting, as of the date our contract ends, which is
October 21st.  Specifically:

1) Drop the Centrex system and return to a POTS system.
2) Drop 761-3554, 761-3596, and 761-7541.
3) Keep 761-300, 761-5041, 761-3411, and 761-3451, and put a hunt on them,
   in that order.

I asked whether there were any features on those lines that we could drop,
line touch-tone dialing.  The answer was that there was nohing we could do
to reduce the base price, because all of their calling plans are geared
toward people who make outgoing calls, and we don't make any. 

I forgot to ask about any kind of circular hunting.  I realized I didn't
quite understand that - is the hunt supposed to give up on a line that's
ringing open after a certain number of rings, and move on to the next one?
That sounds fancier than what a POTS system can handle.  My gut feeling
is that we'd already given them enough to do anyway, and the chances are
already high that they will screw something up.  So let's wait and get
this change taken care of, and then if we want to add something else, do
it later.

The changes should take place on the morning of Tuesday the 21st.  We may
experience a disruption in service, but hopefully it will be short.

response 46 of 95:

Oct 6 20:54 UTC 2003

OK, I get it.  We should be back to the old system now.  Since there are
only 4 lines, it should be easier than before to cycle through them
manually, if necessary.

response 48 of 95:

Oct 7 03:41 UTC 2003

Yes.