[MUD-Dev] RE: Ethernet NICS, maximum connections..mud testing.

[Ben Greear]

I built a stress tester for my game today..using our good friend
java.

No hard figures here..but I'll offer general vagaries!

I have one $12 NIC, and a more expensive NIC (3COM), and
hub and cables connecting these two computers...

I logged on around 100 characters (Interesting note:  the tester
brought my faster machine to it's knees before my mud on my
other machine faltered!).  Each character ran one of 4 simple
scripts in loops.  In about 10 minutes, each of the 100 testers
had logged around 30 K of messages, on the client side.  This
gives an idea of the through-put involved.

On my MUD machine, (I'm running my java client, Xemacs, X, and
Netscape, and of course my MUD).  At full bore, my mud was at around
80% CPU, system at around 12% CPU (This is the NIC, I assume.)

My client tester machine was running 4 java instances (gotta get around
the open socket max) and top, as well as Xemacs and X.  It was maxxed
out, with the system requiring around 12% as well.



My conclusions:  It aint the NIC that will kill you.  I need to
optimize some server code.  Java on Linux is still real slow!

Would anyone be interested in my java tester program?  I think it could
be modified easily to work with most any MUD, providing you know something
about programming and or java.

If anyone wants it, I'll slap in some comments and package it up
nicely for ya...just let me know.

Ben Greear (greear at cyberhighway.net)  http://www.primenet.com/~greear 
Author of ScryMUD:  mud.primenet.com 4444
http://www.primenet.com/~greear/ScryMUD/scry.html