geoff@eagle_snax.UUCP ( R.H. coast near the top) (11/17/88)
In article <400@hafro.UUCP> brb@hafro.UUCP (Bjorn R. Bjornsson) writes: >A friend wants to know how many PCs can reasonably be >handled by a single PC-NFS server, given that all the >PCs are active doing light to medium editing and programming. > >Assume a Sun-3/50, Sun-3/60 or equivalent server and vanilla >PCs. > >Does anybody know? How many PCs have you seen hanging off >a single server? > > Bjorn R. Bjornsson > brb@hafro.is We're actually in the process of trying to gather this data via the Sun field engineering organization. One problem is that in the studies we and a couple of our customers have done we never seem to hit the (expected) "knee" in the (load/response) curve. Obviously a lot depends on application profile, amount of sharing of data, whether or not the Sun is being used for any other work, type of Ethernet card used in the PC, how noisy the network is (lots of broadcasts tend to do a number on 3C501s :-( ), etc. Statelessness obviously helps: we don't seem to run into any limits based on server data structure sizing. I do know that several customers have hung 30-40 PCs off a 3/60 without ill effect. Any and all war stories gratefully accepted! Geoff -- Geoff Arnold, Sun Microsystems Inc. +------------------------------------+ PC Distributed Systems(home of PC-NFS)|Someone, somewhere, wants an RFC822 | UUCP: {hplabs,decwrl...}!sun!garnold |message from YOU. | ARPA: garnold@sun.com +------------------------------------+