On 23 Aug 2004 at 16:46, Eric Dannewitz wrote:

> David W. Fenton wrote:
> 
> >. . . Fragmentation is a big issue. . . .
> >  
> >Not even close to being as important as disk caching, since it only
> >matters on initial read of the file, and really only matters with
> >large files.
> >
> >>. . . I know my
> >>Windows XP machine has to get defragmented every couple of days, . .
> >
> >???
> >
> >NTFS does basic background defragmenting (it doesn't rewrite the
> >files into completely contiguous blocks, but does try to get the data
> > pages into an order that is more efficient for the drive to read
> >than the default assignment off the free chain).
>
> Totally not true. . . .

Sorry, but it *is* true.

> . . . I see it all the time on my XP Machine. I have to
> run the Windows Defragmenter or Disk Keeper all the time. And I don't
> do anything other than some basic internet, QuickBooks, and Finale.
> And the speed improvements are very noticeable. Plus every other week
> we have some sort of Windows patch that needs to be
> downloaded.......which doesn't do anything favorable for the state of
> fragmentation.

You have something wrong with your machine that is not common on 
other machines. At least, I've never seen anything like what you 
describe. Ever. On literally dozens of machines that I'm responsible 
for.

> >I hardly ever defrag any of the machines I'm responsible for, and
> >it's never improved performance noticeably when I've done it. Why?
> >Because defragging basically is only relevant for large files that
> >are read often. And once read, they are cached in memory, so if the
> >original file image is fragmented, it really doesn't matter once it's
> > been read into memory.
> 
> No, it is not. If your drive looks like swiss cheese and the HD is
> going all over the place to get files, then it matters. It's funny
> that you seem to know a lot, but don't really see that defragmentation
> does help with improved performance.

I didn't say it doesn't help at all. I said it doesn't help enough to 
be noticeable.

This is based on my experience with dozens of machines I've been 
responsible for over many, many years, and all different versions of 
Windows.

> >I don't think fragmentation matters for temp files.
> >  
> >
> Really? Have you tried it? I have, and I do notice improvement.

None of my temp files are large enough for fragmentation to matter. 
Second, since they are in constant use, read from and written to, the 
ordering of the data pages on the drive doesn't really matter, 
because they are in RAM, in the disk cache. Reading data from there 
is going to be much faster than either fragmented or unfragemented 
disk image.

> >I disagree entirely. My bet is that all those files are in the disk
> >cache, not being read/written from/to the disk each time (actually,
> >an explicit save probably does a forced write to the original file,
> >but changes in the temp files themselves are probably in the write-
> >ahead cache, and reads are coming from the cache in most cases).
>
> I don't think so. Perhaps one of the Finale tech guys on the list can
> explain it. I would be interested to know exactly what happens, not
> speculation about what happens.....

This has nothing to do with Finale -- it's all about how the Windows 
disk caching works. Files are read from the disk can cached so that 
they don't have to be re-read from the disk.

That has nothing to do with Finale.

And once in RAM, it doesn't matter whether the file image on disk was 
fragmented or not.

> >>. . . I see my little hard drive light go on all the
> >>time with Finale. . . .
> >
> >That doesn't mean anything. OS subsystems may be doing any number of
> >things that cause disk reads/writes.
> 
> But this harks back to the issue of fragmentation. If your drive is
> fragmented, it does take a toll on your performance. Doesn't matter,
> Windows or Mac.

A small toll, one that is worth defragmenting every month or so or 
after any major churning of your drive (such as an OS upgrade). Also, 
one thing that makes a huge difference is how you partition your 
drives. If you have a single drive partition for OS and for programs 
and for data, then you're going to have more problems with 
fragmentation than if you have a partition for your OS, a separate 
partition for your programs and a third for data. You shouldn't have 
to defrag the system partition very often at all, except after an OS 
patch or a major upgrade. The programs partitition shouldn't need to 
be defragmented except after you've installed a bunch of new 
applications (program folders are mostly read-only, so once 
defragged, they stay that way). The data partition could benefit from 
being frequently defragged if your data files are quite large (on the 
order of 10s of megabytes, e.g., graphics files).

But I've never seen a system's performance improve dramatically after 
a defrag, unless the machine was very badly defragmented and had not 
been defragged before. Regularly defragging, while a good idea, 
should not result in noticeable performance increases, unless, 
perhaps, your machine is marginal in terms of CPU speed and RAM in 
the first place. With lots of RAM, you won't see much in the way of 
improvement with frequent defragging.

> >I stopped worrying about it a long time ago when it became clear to
> >me that MS had engineered something the worked extremely well and so
> >there was no need to try to tweak it any more.
> 
> So I suppose utilities like Disk Keeper (http://www.diskeeper.com/)
> then are totally useless. I can't imagine running my Windows machine
> without Diskeeper.....

Most utilities are worthless except for data recovery and disk 
repair. My experience with most utilities is that they make you feel 
they are worthwhile by falsely informing you of how many problems you 
have. I've often seen utilities that run as background monitoring 
processes cause most of the problems they are reporting.

Regular maintenance with the tools provided with the OS are 
sufficient for most problems.

-- 
David W. Fenton                        http://www.bway.net/~dfenton
David Fenton Associates                http://www.bway.net/~dfassoc

_______________________________________________
Finale mailing list
[EMAIL PROTECTED]
http://lists.shsu.edu/mailman/listinfo/finale

Reply via email to