home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.periphs.scsi,comp.sys.sun.hardware,comp.sys.sgi.hardware,comp.sys.next.hardware,comp.sys.mac.hardware,comp.sys.ibm.ps2.hardware,comp.sys.ibm.pc.hardware.storage,comp.sys.ibm.pc.hardware.misc,comp.sys.ibm.pc.hardware,comp.sys.hp.hardware,comp.sys.amiga.hardware,comp.os.linux.hardware,biz.comp.hardware
- Path: netcom.com!stephenk
- From: stephenk@netcom.com (Stephen Knilans)
- Subject: Re: HELP: Testing a Quantum SCSI harddisk.
- Message-ID: <stephenkDpuF8K.M79@netcom.com>
- Organization: Netcom Online Communications Services (408-241-9760 login: guest)
- References: <50077815@wbmt41.wbmt.tudelft.nl> <4jnrts$qum@gazette.engr.sgi.com> <Dp9GB7.F2q@avalon.chinalake.navy.mil>
- Date: Sun, 14 Apr 1996 08:21:56 GMT
- Sender: stephenk@netcom.netcom.com
-
- In article <Dp9GB7.F2q@avalon.chinalake.navy.mil> Dave Bower <bower> writes:
- >Disks from various vendors have been voodoo to me. I have encountered
- >numerous disks from different vendors (hp, seagate, quantum, DEC) that have
- >suddenly exhibited degraded performance. I've tried using SYSDIAG on them and
- >talking to the Response Center. No luck. They fsck just fine. The only way I
- >can tell they are hurting is that they respond slowly in a script I wrote to
- >simply write and delete files of various sizes. When a disk shows degraded
- >performance, it's typically 5 times slower than a well-behaved disk. Over the
- >past 3 years, I've had about 10 disks "become slow". Any ideas as to how to
- >diagnose the problem?
- >
-
- MOST filesystems can get VERY badly fragmented, and this can REALLY slow things
- down! How much? A benchmark I wrote indicates thata disk can slow down to
- 1/60th the speed! In other words, if it takes 1 second to load a 1 megabyte
- file, then it COULD(if 100% fragmented) take up to 1 MINUTE to load!
- Although that is the ABSOLUTE worst case(assuming a HALFWAY decent system),
- I HAVE seen programs slow down a LOT! At work, for example, some compiles
- that USED to take about 20seconds now take 3.5 MINUTES(or over 11 times as
- long)!
-
- How do you solve it? The BEST way is through a defrag program. There is one
- for Linux, but I don't think it works. I MUST say, though, that the latest
- version of fsck claims that my system fragmentation is LESS than 8%! That is
- AMAZING! I have used it for about 2 years or so as it is, and CONSTANTLY
- update, etc.... It has FAR more reason to be fragmented than my system at
- work, yet it ISN'T! GOOD GOING LINUS!!!!!!! 8-)
-
- Dos and Stacker both come with defrag programs.
-
- Steve
-
- BTW this is NOT a disk problem! It is a filesystem program. The ONLY system
- I have ever worked on that couldn't have fragmentation is RT11 and TSX-11!
- Why don't the others take a cue from IT? I didn't say they DIDN'T have
- fragmentation, but that they COULDN'T! If you had 50 holes that EACH had 1MB
- of disk space, the largest file you could have would be 1MB! WHAT do you
- do if you want to use all 50MB? EASY! You simply dismount the pack, and
- SQUEEZE it! ANY bad blocks, or power failure, or system crash, and the pack
- is WORTHLESS! After this, you could save a 25MB file! WHY? Because it has
- NO idea how to use space(based on possible future use). Thus, it would
- use half the largest available space, or all of the second largest space
- available, which ever is greater. If it allocated block by block, and
- someone else used the space, you would be DEAD!
-
- So, we have to settle for slower performance in order to have the
- flexibility and safety.
-
-