Return to BSD News archive
Xref: sserve comp.os.386bsd.questions:7976 comp.windows.x.i386unix:6543 Path: sserve!newshost.anu.edu.au!munnari.oz.au!foxhound.dsto.gov.au!fang.dsto.gov.au!myall.awadi.com.au!myall!blymn From: blymn@awadi.com.au (Brett Lymn (master of the siren)) Newsgroups: comp.os.386bsd.questions,comp.windows.x.i386unix Subject: Re: If you were to assemble a new machine... Date: 15 Jan 1994 13:19:05 GMT Organization: AWA Defence Industries Pty. Ltd. Lines: 54 Message-ID: <BLYMN.94Jan15234905@siren.awadi.com.au> References: <1994Jan12.161313@unccsun.uncc.edu> <crt.758474839@tiamat.umd.umich.edu> <2h49e2$fde@homer.cs.mcgill.ca> NNTP-Posting-Host: siren.awadi.com.au In-reply-to: storm@cs.mcgill.ca's message of 13 Jan 1994 20:03:14 GMT >>>>> "Marc" == Marc WANDSCHNEIDER <storm@cs.mcgill.ca> writes: In article <2h49e2$fde@homer.cs.mcgill.ca> storm@cs.mcgill.ca (Marc WANDSCHNEIDER) writes: Marc> In article <crt.758474839@tiamat.umd.umich.edu>, Rob Shady Marc> <crt@tiamat.umd.umich.edu> wrote: >> My personal favorite that falls into this price range would be >> something similar to the following: >> [shopping list deleted.] >> Seagate 9GB SCSI Drive $3500 Marc> actually, i really don't like the disk selection there. I dunno 9 Gig sounds quite nice to me... Marc> last i heard, there was a limit of 8 partitions on any given Marc> disk. that means, after you lose swap [let's be generous, and Marc> say it's 1gb], the c and d partitions for dos stuff you've still Marc> got to spread only 5 partitions with 8gb of space. Your numbers do not quite add up... you forgot to allow some space for the dos stuff, or maybe that was intentional ;-) Marc> that means quite a few partitions that are obscenely large. The only real problem with 1 Gig partitions is that it takes a dim age to fsck them. I have had a 1Gig partition on my machine starting in 0.1 386bsd days and now with NetBSD 0.9 without any problems. The fragmentation of my large partitions is of the same order as my other partitions. Marc> there are probably some perfromance impacts on this. There are some performance problems having all your storage on the one spindle, it restricts the data path to the oxide. I missed the original post so I cannot say if this will really be a problem in this case. Marc> i'd rather have a pile of smaller disks. The problem with lots of disks is that you MTBF goes way down, the more pieces you have the more likely one of them is going to fail. Of course having a single disk will mean you are in deep do-do if it fails where as with lots of disks you may be able to work around things, but the single big disk is, statistically, less likely to fail. A lot of small disks would be good if you are looking for fast disk performance by using striping or one of the other RAID mechanisms. -- Brett Lymn