home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.unix.wizards:5229 comp.unix.admin:6645
- Newsgroups: comp.unix.wizards,comp.unix.admin
- Path: sparky!uunet!cis.ohio-state.edu!magnus.acs.ohio-state.edu!usenet.ins.cwru.edu!agate!stanford.edu!rock!concert!uvaarpa!murdoch!brain.med.virginia.edu!prs9k
- From: prs9k@brain.med.virginia.edu (Phil Scarr)
- Subject: Re: Causes Of Death On Heavily-Loaded UNIX System?
- Message-ID: <1992Dec15.052914.14409@murdoch.acc.Virginia.EDU>
- Sender: usenet@murdoch.acc.Virginia.EDU
- Organization: Neuroclinical Trials Center, University of Virginia
- References: <1gjlnjINN9eb@smurf.sti.com>
- Date: Tue, 15 Dec 1992 05:29:14 GMT
- Lines: 20
-
- Much of this is raw interrupt delay. The interrupt handler and message
- handler simply get overwealmed. What you are doing is equivalent to
- running OS/2 on a 4mb 386sx16mhz and asking why it is slow. The system
- performance. How much RAM does the system have? How many people are
- running Xwindows? Remember, each user shell requires a minimum ammount
- of real (i.e. non-virtual) memory. If you are pushing your Sun this
- hard and you don't have enough memory, you could run into some REAL
- problems with swapping delays. The system is spending a lot of time
- figuring out what it can and cannot swap to disk at any given time.
-
- One simple fix might be to add lots more memory to the box (if you can).
-
- -Phil
- --
- PHIL SCARR \ We are Microsoft... / (o) 804.243.0229
- University of / OS/2 is irrelevant. \ (f) 804.243.0290
- Virginia, \ UNIX is irrelevant. / prs9k@Virginia.EDU
- Neurosurgery / Openness is futile. \ prs9k@Virginia.BITNET
- HP-UX is my life!\ Prepare to be assimilated... / ...uunet!virginia!prs9k
-
-