| T.R | Title | User | Personal Name
 | Date | Lines | 
|---|
| 909.1 |  | MARVIN::TURNER | Neil Turner IPEG REO, 830-4140 | Wed May 21 1997 03:57 | 9 | 
|  | What problems are you seeing, you do not make it clear whether
you are seeing problems or just anticipate them?
The default retry timer on PPP is 3000mS which should be fine
with your 250mS delay. This timer is only used whilst the
link is initialising, once up PPP is an unacknowledged datalink
so I don't envisage and problems over your connection.
Neil
 | 
| 909.2 | satellite link up and running.Performance ? | BARNA::DSMAIL |  | Fri May 23 1997 10:55 | 39 | 
|  | 
	Thanks for your quick response.
	
	Really I wanted to advance to the problems because I have no 
	experience about satellite links and I was afraid about link
	establishment and its performance.
	Well , now I know a little more and I tested the configuration
	in the customer site , but I had some performance problems.
	Following explains you the  parameter configuration and its results
	With PPP configuration parameters:
        Maximum frame size=2048 Bytes (default) making ftp with a file size 3MB
	the transfer speed arrives to 20Kbps aprox.
	Maximum frame size=512  Bytes (default) doing a ftp with the same file size
	the speed becomes to 14 Kbps.
	
	Maximum frame size=4088 Bytes (maximun ) the tranfer speed arrives to
	28 Kbps .
	With max frame size 4088 Bytes and CCP compression enabled we reached the 
	max speed 42 Kbps.
	Note that the channel speed is 64 kbps .
	Doing a ping  the response time was 
	Packet size 64 bytes ------- 550 ms
	Packet size 4000 bytes ----- 1600 ms aprox.
	What do you think about ? 
Thanks in advance
Luis Miguel Sanchez
 | 
| 909.3 | TCP window size? | MARVIN::TURNER | Neil Turner IPEG REO, 830-4140 | Thu May 29 1997 02:52 | 9 | 
|  | I don't think your problem lies with PPP since as I said before it
is an unaknowledged datalink protocol. However I am not sure what TCP 
will do when it is used on a link which has such a long round trip delay
(550mS in your case). It is going to need a large window size in order 
'fill up' this delay and therefore saturate the line. I don't know enough 
about the behaviour of TCP, but I suspect that its default window size 
will not be large enough to make best use of the line.
Neil
 |