Carnagey, Bob/NTSEUG ([log in to unmask]) wrote:
: We have just installed a frame relay network between our US site and several
: other sub offices throughout the world.   While the system is working fine,
: we have a problem with latency, the delay caused by the HP3000 requirement
: for "echoing" back characters to the users screen.  What this means is that
  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  - The only time this would occur would be if the application disabled
    echoing as it usually is done by the DTC port without any host
    interaction and the DTC only sends data to the host at read completion,
    error or byte count exhausted.  Of course, with an application doing
    one byte reads and then echoing it back one would see a host interaction.
 
 
: it takes, on the average, about 280 ms for a character to loop from the
: users keyboard, across the frame relay network to the HP3000 and for the
: HP3000 to "echo" back to their screen the character.
 
  - What kind of terminals do the users have?  Dumb terminals?  Connected
    to a DTC (on HP3000 side)?  Or is this a VT-connection from a PC?  Is
    the application doing one character reads?  The reason I'm asking all
    these questions is that I don't fully understand how your users are
    connected.  If you have a frame relay network,  I take it that your
    terminal users cannot be connected to it directly,  there is some asynch
    port somewhere in the picture (unless it's VT and PC's attached to LAN).
    Anyway, since you indicated that it used to be fine with X.25 PAD's I
    assume that your users have dumb terminals(?)...
 
 
    HP3000
      |
   o--+------+---o
           __|___     (                   )
          |router|__(   Frame relay network )--- ????????? --- terminals???
          |__?___|   (                     )
 
    What else is there in this picture??  ...a graphic description would be
    very helpful...
 
 
:                                                      In our previous X.25
: network, we had the PAD do the echo thus making it look like the HP3000 had
: done it.  Since the PADs were at each site, there was no perceptible delay
: for the echo.
 
  - With PAD's (which cannot terminate a read on a byte count), this probably
    has been something (and now this is all guesswork) that despite of the
    application doing short reads, the PAD would still echo locally and
    transmit to the host only depending on how the PAD was configured.
    Either it would have forwarded data on idle timer expiring (an X.3 parm),
    or whatever the data forwarding character was set to (maybe even all?).
    However,  since I don't have good enough understanding on how these two
    different configurations are/were exactly set up,  I'm afraid I cannot
    give you a good answer.
 
: Does anyone have any ideas on how to do this with a router or any alternate
: method?  We are using Xyplex routers.
 
: One other issue, we are using telenet to connect these users to the HP3000.
:  They are experiencing periodic telenet disconnects and there is no obvious
: reason why.   Has anyone seen this type of problem before?
 
  - Sorry my ignorance -- I don't know what telenet is and how does it work
    (I take it's an US thing? - my background is in Europe) - someone cares
    to explain.
 
 
Cheers  :-)
Eero Laurila - HP CSY Networking lab, NS services.