On Thu, 2004-11-11 at 10:44, Marvin T. Pascual wrote: > Server: IBM xSeries 345 > Processor: Dual Xeon 2.8Ghz > Memory: 4GB RAM > Hard Disk: 2 x 73GB SCSI > Workstations: Intel Pentium 1 to 4 with 100Mbps NICs > Switches: 3Com SuperStack 10/100Mbps > Applications: OpenOffice.org, Novell Evolution, Mozilla Firefox, etc.
30 - 40 would be my vague guess, limited by RAM rather than CPU. You're dealing with a lot of memory-hungry apps, though. In particular, I find Evolution to gobble mind-boggling amounts of memory (at least when using IMAP). I have a dual Xeon 2.4GHz with 2GB of RAM here and it's happy with twelve workstations. The CPU is essentially idle all the time - the pressure is all on RAM. Memory use is at about fifty percent, but we run a mail server (SpamAssassin is an incredible memory hog), web server, and a few small databases off the box too. Most of ours users use XFCE4 with Mozilla and OO.o. Three use KDE 3.3 (size-optimised) with Evolution 1.4, OO.o and Konqueror. The biggest memory hog is Evolution, by far. I'm hoping Evolution 2.0 solves that problem. OO.o is next, then Mozilla. KDE is barely detectable when it comes to memory use, as it's per user data set sizes are amazingly tiny. If you can persuade OO.o and Evo to gobble less RAM, you may well be able to more than double my user count estimate. I'd strongly recommend that you set up your software environment and about three terminals, then start profiling memory use. Look at how much extra memory use the first, second, then third terminals coming online causes (there should be no X server on the server for this test). The second and third terminals should cause roughly the same jump in memory use on the server, and the amount they've been using after they've been running for a while is probably what you want to plan for per-terminal. Remember that for optimum efficiency you'll want at least 512MB free for a monster disk cache. If our server gets below that we start touching the disk, and things slow down significantly. Of course, this is in the end just my personal opinion, and it's mostly guesswork based off my experience running a similar system in a small environment. Others may have been much more successful or may think my estimates are overly optimistic. -- Craig Ringer ------------------------------------------------------- This SF.Net email is sponsored by: Sybase ASE Linux Express Edition - download now for FREE LinuxWorld Reader's Choice Award Winner for best database on Linux. http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click _____________________________________________________________________ Ltsp-discuss mailing list. To un-subscribe, or change prefs, goto: https://lists.sourceforge.net/lists/listinfo/ltsp-discuss For additional LTSP help, try #ltsp channel on irc.freenode.net