home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.aix
- Path: sparky!uunet!gatech!bloom-beacon!bloom-picayune.mit.edu!athena.mit.edu!jfc
- From: jfc@athena.mit.edu (John F Carr)
- Subject: Re: JFS i-nodes summary
- Message-ID: <1992Aug31.001843.16315@athena.mit.edu>
- Keywords: jfs i-nodes filsystem
- Sender: news@athena.mit.edu (News system)
- Nntp-Posting-Host: achates.mit.edu
- Organization: Massachusetts Institute of Technology
- References: <1992Aug28.174656.8854@fcom.cc.utah.edu>
- Date: Mon, 31 Aug 1992 00:18:43 GMT
- Lines: 27
-
- In article <1992Aug28.174656.8854@fcom.cc.utah.edu>
- ben@osiris.usi.utah.edu (Ben Pratt) writes:
-
- > We figure
- > that, given our situation and needs, that is a waste of 55 to 60
- > meg per 2 gig filesystem. Since we have 8 of them, that means we
- > are losing 440 Megabytes to 480 Megabytes of scratch file space
- > for our applications... Ouch.
-
- 55 / 2000 = 2.8%
-
- A loss of less than 3% of space merits an "ouch"? If you're normally
- running at 97% capacity you've got bigger problems than JFS.
-
- Have you investigated the performance of JFS as you approach 100% use?
- You may be better off not using the space. The Berkeley fast file system
- reserves 10% in addition to the inode space to prevent performance
- degradation, but I rarely hear any complaints about inefficient use of
- space.
-
- If you can get down to a small number of files of nearly constant size,
- you should not use JFS and instead use an unformatted disk partition
- (/dev/something). This will have better performance and allow you to use
- more space.
-
- --
- John Carr (jfc@athena.mit.edu)
-