Modem-to-modem flow control (was Re: The "Final" Word on 3b1 HFC)

[Floyd Davidson]

In article <1991May19.190618.14548 at netcom.COM> gandrews at netcom.COM (Greg Andrews) writes:
>
>Hello.  I'm here to tell you that it WILL when error correction is active.
>It doesn't send HFC signals, to be sure.  But it does send flow control
>signals ("messages" is a better term) through the error correction protocol.
>

It won't send HFC signals, which is what I said.

>>
>>Even at 2400 bps between modems one could still have a problem if the

The "problem" I was talking about is losing data.

>>modem to 3b1 rate is 19.2 kbps if the modem buffers receive data and
>>                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>dumps it to the 3b1 at a full 19.2 kbps rate.
>>^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>Think about it.
>>
>
>Again, when error correction is active, you'll get exactly that.  The data
>would come across the phone line in a packet, and the receiving modem can't
>release that data until the whole packet arrives intact.  As soon as the
>modem checks the packet for errors, it releases the data to the computer at
>the full speed of the RS232 interface.  19.2 Kbps, in your example.
>
>Thus the computer receives data at an AVERAGE of 2400 bps, but the computer's
>ability to handle the data rests on how it can handle data in bursts as well
>as over the average.  With error correction active, the modem would feed 
>data to the computer in full speed bursts with pauses between each burst.  
>The amount of data in each burst will depend on the data flow and which error 
>correction protocol is being used.  MNP4 can use up to 256 byte packets, with 
>MNP5 compression increasing that up to about 500 bytes.  V.42 can use packets 
>up to 4096 bytes in length, with V.42bis boosting that up to 16,000 bytes.  
>
>Of course, the REAL data burst lengths after decompression will vary, and 
>probably won't be as large as the numbers I've stated here.  My point is that
>the receiving computer will receive bursts of data at full speed, and those
>bursts can be fairly large in size.  If the computer can't handle that much
>data in one burst, then it will drop data on the floor and 
>your statement will be untrue.
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

My statement was that it would lose the data, which as you've pointed
out is true.

Thanks for posting Greg. My first post on this suggested that it
be taken to comp.dcom.modems and you are specifically the person
who posts to that group that I was thinking of...

Floyd
-- 
Floyd L. Davidson   | Alascom, Inc. pays me, |UA Fairbanks Institute of Marine
floyd at ims.alaska.edu| but not for opinions.  |Science suffers me as a guest.