dial-in lines being held for "days"

[Jeff.Ogden]

Attached is a long string of messages talking about busy signals
on the 9600 bps dial-in lines in Ann Arbor and the fact that at least
some of the problem is caused by people holding these lines for
very long periods of time.
 
I'd like to start a discussion about what we can and should do about
this. It is my hope that we can find a common policy for all of the
"Michnet" 9600 bps dial-in.
 
Things I'd like to know include:
   --are schools other than UofM seeing similar problems?
   --is this mostly a problem on the 9600 bps lines (the ones
     that support SLIP and PPP) or is it a problem on the
     "19.2K" lines too?  The slower speed lines?
   --do the 9600 modems that other schools have all support the
     inactivity timer?
   --do you think that an inactivity timer is something we should
     try?  If so, how long? 30 minutes? 45 minutes? ...
   --will an inactivity timer work or is there enough housekeeping
     traffic on PPP and SLIP connections to make an idle line seem
     busy from the modems point of view?
   --there has been some talk about making changes to the PCPs and
     SCPs to have a four hour or a six hour timer on public dial-in
     lines that would drop the lines even if there were activity.
     That is a big step and would cause problems for some users
     who are doing real work. Still a single four or six hour
     session is a long one and not all that common and if we are
     giving busy signals to people forcing people off is a way to
     share the busy signals.
 
Comments, suggestions and thoughts welcome.
 
  -Jeff
 
 _________________________________________________________________
 
 Message: 15157508, 7 lines
 Posted: 10:42pm EDT, Wed Aug 12/92, imported: 10:40pm EDT, Wed Aug 12/92
 Subject: A2 busy signals
 To: Jeff Ogden
 From: rgs@merit.edu
 
 I realize that Scott and Fred are really the ones I should be telling
 this, but anyway ...  I just got busy signals on both the 9.6 and 19.2
 lines (and am on at 2400 currently).  There are 10 9.6 lines in use,
 and 13 in the 19.2 sequence, and the kiddies aren't back yet.
 
   - Rick [Schmalgemeier]
 .................................................................
 
 Message: 15164061, Reply to: 15160054, 7 lines
 Posted: 10:38am EDT, Thu Aug 13/92
 Subject: A2 busy signals
 To: Jeff Ogden, Russell J Dwarshuis, mike bowdish, W. Scott Gerstenberger
 From: Fred Rowe
 
 We are in the process of ordering and installing 16 more 9.6/19.2
 in some combination.
 
 MY GUESS IS THAT WHEN RUSSELL GETS THESE 16 INSTALLED WE WILL
 STILL GET BUSYs. I WILL CONTINUE TO INSTALL THEM UNTIL I RUN
 OUT OF BUDGET..
 FRED
 .................................................................
 
 Message: 15176948, Reply to: 15175049, 3 lines
 Posted: 9:02pm EDT, Thu Aug 13/92
 Subject: A2 busy signals
 To: Jeff Ogden
 From: Mike Alexander
 
 Thanks.  For what it's worth, my experience is that almost all the
 busy signals are on 747-3400 [9.6] I almost always get in on the first
 try on 763-6800 [19.2].
 .................................................................
 
 Message: 15181376, Forwarded message: 15176948, 6 lines
 Posted: 8:59am EDT, Fri Aug 14/92
 Subject: A2 busy signals
 To: Fred Rowe, Russell J Dwarshuis, mike bowdish
 Cc: W. Scott Gerstenberger
 From: Jeff Ogden
 
 FYI - Mike A's comments would seem to support the idea that it is
 SLIP/PPP users rather than regular async users that are causing the
 increase in use.  I have gotten a busy now and again at 19.2K
 however.  What are your observations on how busy the 9.6 vs 19.2
 lines are?
    -Jeff
 _________________________________________________________________
 
 Message: 15171906, 14 lines
 Posted: 3:28pm EDT, Thu Aug 13/92
 Subject: 9.6 use / abuse
 To: Russell J Dwarshuis, Glenn S. Wiltse
 From: mike bowdish
 
 I know people are bitching about the 9.6 modems being busy all the time, but
 i think something has changed dramatically in the way they are being used in
 the last couple of months. It looks like the people who get the SLIP, PPP type
 connections, stay on forever, I just looked at the opsink and only one modem
 (approx.) is making connections on and off. We need to take a look at the
 activity of DSF1 to see what is going on with the other ports. I know these
 modems are free for the public to use, but if somebody is staying on for a
 few days at a time, or a week or whatever, I would consider that a little
 bit abusive. Bert has mentioned in the past, that he thought somebody has
 been staying on as long as they can for some sort of cheap IP connection.
 Glenn, can we make this a agenda item for the meeting tomorow ?
 
 . . .
 
  MB
 .................................................................
 
 Message: 15179808, Reply to: 15171906, 6 lines
 Posted: 7:42am EDT, Fri Aug 14/92
 Subject: 9.6 use / abuse
 To: mike bowdish, Russell J Dwarshuis, Hardware Support, W. Scott
     Gerstenberger, Jeff Ogden
 From: Glenn S. Wiltse
 
    Well Mike, I asked Fred about this and he doesn't feel it would be of much
 benifit to talk about this at our shop staff meeting. Aparently this is already
 being discussed at higher levels. I feel much the same as you, allowing no time
 restrictions on these types of connections doesn't seem like a very good thing.
   I've added Scott and Jeff to the recipient list of this message, perhaps
 one or both of them could comment on any policy in regards to this.
 .................................................................
 
 Message: 15180570, Reply to: 15179808, 10 lines
 Posted: 8:28am EDT, Fri Aug 14/92
 Subject: 9.6 use / abuse
 To: Glenn S. Wiltse, mike bowdish, Russell J Dwarshuis, Hardware Support,
     W. Scott Gerstenberger, acr@merit.edu, jrv@merit.edu
 Cc: Eric M. Aupperle
 From: Jeff Ogden
 
 Glenn and Mike,
   I agree that people holding dial-in lines for long periods is a problem.
 The questions is what can we do about it?  If we come up with a policy what
 would be a good one?  In addition to a policy, how would we enforce it?
 Is enforcement required? Can it be done manually or does it need to be
 built into the network in some fashion? In the past people have talked about
 charging by the minute for dial-in to get people to get off when they aren't
 really doing work. I don't think the other member schools would go for that
 solution.  Suggestions welcome.
    -Jeff
 _________________________________________________________________
 
 Message: 15195975, 12 lines
 Posted: 9:53pm EDT, Fri Aug 14/92
 Subject: busy signals ...
 To: Fred Rowe
 Cc: W. Scott Gerstenberger
 From: Jeff Ogden
 
 Fred,
   With the recent messages about very long calls particularly at 9600 and
 19.2K I'd like to see if we could gather some counts that would give us
 a feel for how big a problem this really is?  Could you have someone do
 something like this:  Every night at 4 am (or any other time that works)
 make a list of connections via the 9600 and 19.2 AA dial-in sequences
 and then compare the connections from the previous day to see how many are
 still on.  After we week or so of this we should get a feel for how serious
 the long holding time problem is?  Need to take some care to list all
 types of connections (async, PPP, SLIP, SLFP).  The Sense all command
 does this I think.
   -Jeff
 _________________________________________________________________
 
 Message: 15260992, 39 lines
 Posted: 3:35pm EDT, Wed Aug 19/92
 Subject: 9600 bps modem use/abuse
 To: Hardware Support, huns@merit.edu
 From: Russell J Dwarshuis
 
 I've recently finished reconfiguring all the 9600 modems so that
 they're set for MNP4 and no data compression or flow control.  I
 freed the modem for removal from the hunt group by busying it out
 via the front panel switch.  This busies out the modem when the
 current user hangs up.  There were four (mis)users who stayed on
 for (at least) two days without disconnecting.  I finally nuked
 them so that I could change the modem out.  I was curious about
 what kind of use the 9600's were getting, so I periodically
 looked for any receive or transmit data from the modems.  Here's
 what I found:
 
 hunt group          8/13               8/14
 number                   time of check
          12  1  2  3  4  5  6     11 12  1  2  3  4
   1       C  C  C  C  C  C  C      C  C  C  C  C  C
   2       C  T  T  T  T  C  C      C  T  T  T  T  C
   3       C  C  C  C  C  C  C      C  C  C  C  C  C
   4       C  C  C  C  C  C  I      C  I  T  T  I  T
   5       T  T  T  C  T  T  C      C  T  T  T  T  C
   6       I  T  T  T  C  T  I      I  I  T  I  C  T
   7       C  C  C  C  C  C  C      C  C  C  C  C  C
   8       C  C  C  C  C  C  C      C  C  C  C  C  C
   9       T  C  T  C  T  T  T      C  T  T  T  C  C
  10       I  T  T  C  C  C  T      C  C  C  I  T  I
 
 "C" means carrier only, no transmt or receive data lights.
 "T" means traffic (transmit and/or receive data lights on).
 "I" means idle (no carrier).
 
 The (non)users on 1, 3, 7, and 8 never did disconnect...
 voluntarily...
 
 We've just added six more 9600 modems to the hunt group.
 I'll check the useage pattern in another couple of days.
 The 9600 and 19200 bps modems that we use have an option
 that can be enabled that will disconnect the user if no
 traffic is received in a (selectable) amount of time.
 This could be one way of getting rid of the dead wood.
 Comments?
 .................................................................
 
 Message: 15268422, Forwarded message: 15260992, 8 lines
 Posted: 7:26am EDT, Thu Aug 20/92
 Subject: 9600 bps modem use/abuse
 To: W. Scott Gerstenberger, Jeff Ogden
 From: Fred Rowe
 
 As you can see from Russells table here 40% of the modem pool was
 tied up for 2 days with no traffic. Not a good use of our limited
 resources.
  I would like to install something like a 30-60 minute time out, then
 reset the modem.
 
 Comments
 fred
 .................................................................
 
 Message: 15268622, Reply to: 15268422, 29 lines
 Posted: 7:46am EDT, Thu Aug 20/92
 Subject: 9600 bps modem use/abuse
 To: Fred Rowe, W. Scott Gerstenberger
 From: Jeff Ogden
 
 If the time out is something that can be done in the modems then I
 think that a 30 minute inactivity timer would be OK.  I think we need
 to "announce" this change in advance. We can do that with something in
 the [UofM] ITD Digest and a short hermes message pointing to a notice in the
 help server.  Do all of our 9600 and 19.2 modems support the time-out
 feature?
 
 Strictly speaking this is a UofM decision rather than a Merit
 decision. However, it would be nice of we could get agreement
 to do something like this from all the Merit schools so it would
 work the same on all the 9600 dial-in.  That will take a little
 more discussion.
 
 We need to wait for Scott to get back and to read through all of
 his e-mail, before making a final decision on this.
 
 My real worry is that the inactivity timer won't help, because
   1, the PPP and SLIP protocols do some background handshaking
      that will be seen as activity by the modems and so prevent the
      time-out (I don't know if this is true and it is something we
      can test),
   2, that it will be easy for the same inconsiderate people who
      hold lines for such long periods to generate enough taffic to
      hold the lines anyway.
 
 Do we need to test the inactivity timer in the modems on one line before
 we take this step?
 
   -Jeff
- - - - - - - - - - - - - - - - -