Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!ucbvax!FTP.COM!backman
From: backman@FTP.COM (Larry Backman)
Newsgroups: comp.protocols.tcp-ip.ibmpc
Subject: Re: Telnet from IBM's TCP/IP for OS/2
Message-ID: <9106231049.AA15512@ftp.com>
Date: 23 Jun 91 10:49:04 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Distribution: world
Organization: The Internet
Lines: 31


 >> 
 >> In article <31500001@acf3.NYU.EDU> chapman@acf3.NYU.EDU (Gary W. Chapman,WWH 318,212-998-3045,718-499-7815) writes:
 >> >I found this (ibm tcp/ip telnet daemon on OS/2 machine) to be a dog 
 >> >on a 20 Mhz model 80 with 12 meg of memory.  We suspect un-optimized
 >> >code...
 >> 
 >> After we found the code to be slow on the 55SX, we tried to use the VT100 Tel-
 >> net client from a PS/2 Model 80-A31, a 25Mhz machine with 14 meg of memory.
 >> 
 >> It was an incredible dog on there also.
 >>


Define Dog; our PCTCP Telnetd for OS/2 is slow in establishing the initial
login prompt (inetd spawns startses which spawns telnetd which spawns
login) but once we are going we get reasonable (~ 9600 baud) performance
on a 386/20 or 25.

If you call it a dog because its slow making its initial connection you might
want to look at your OS/2 configuration.  OS/2 1.3 in particular is
notorious ( 20 seconds on a 4 Meg 386/20) for taking its time in spawning
a child process out of a background program (like inetd).  Also; it
gives emphasis to the foreground app at the expense of background stuff.

If your interested I can dig up an archived tip from comp.os.os2 on how
to optimize OS/2 for better background processing.

Larry Backman
backman@ftp.com