Re: [Linux-decnet-user] naive LAT question regarding speed

[Patrick C. <pa...@ty...>]

Zube wrote:
> On Fri Aug 19 07:45:19 AM, Patrick Caulfield wrote:
> 
> 
>>Could you tell me more about what you are doing? I know that "reverse" 
>>LAT is slower by about a factor of 1.8 or so but I've never heard of 
>>those sort of orders of magnitude. Normal terminal access, in my 
>>experience, is limited by the speed of the serial line between the 
>>terminal and the server.
> 
> 
> Thanks for getting back to me.
> 
> The setup is a few dumb terminals (and a couple of scanners attached to
> dumb terminals) hooked to a DecServer 200.  The logins on the Debian
> Thinkpad machine automatically rlogin to another machine where the programs
> are run, but I don't believe the problem is between the Thinkpad and
> the remote machine.
> 
> For example, when I do:
> 
> c linux 
> 
> from the local prompt, I immediately get a login prompt.  But for a three
> character account, say abc, typed immediately after the prompt appears, 
> the sequence appears this way:
> 
> ab <delay> c
> 
> Another example:  There is a program that is "downloaded" to a scanner
> attached to a dumb terminal.  The code is displayed on the screen as
> it is processed by the scanner.  When the DECStation was in place, the
> code scrolled by at a regular, if unimpressive, clip; now, it scrolls
> at the same unimpressive clip for a few lines and then stops for several
> seconds, then continues for a few more lines and stops, etc.  It all works,
> but the hangs make it tough to work with.
> 
> Perhaps I will try to put Debian + the lat software on a different machine.
> I've seen stranger things occur with Thinkpads; it was just a very 
> convenient choice.
> 

Hmm, that sounds very odd. I use LAT regularly and have never seen that 
sort of slowdown and I run my connections at speeds faster than a DS200 
can even dream of !

Check that the DS200 has the latest firmware installed, I did have some 
problems with old versions. Unfortunately my DS200 has now died and I 
can't remember the version I was using - I do still have the download 
image though :-)

It might also be worth checking that there isn't execssive packet loss - 
I suspect my code that copes with that may not be optimal. It would be 
interesting to see if similar problems occur on another Linux machine.

Patrick