Re: Create Big size dummy File..

Tech-Archive recommends: Repair Windows Errors & Optimize Windows Performance

From: Tony Proctor (tony_proctor_at_aimtechnology_NOSPAM_.com)
Date: 06/24/04


Date: Thu, 24 Jun 2004 13:01:41 +0100

In principle, seeking to a very high file address (i.e. beyond the EOF) and
writing more data there should leave behind an expanse of random data from
the disk - I've certainly used some UNIX systems in the past that have
exhibited this, and very interesting it was to read through it too.

However, I'm wondering if the system file manager is having an influence
here. I know that it uses memory-mapped sections to implement its caching.
Uninitialised pages would then be created as 'demand zero', with the result
that when they're flushed, NULLs would be written to the output file. This
also explains why 'creation' is so much quicker than 'copying': Demand-zero
pages are very cheap to create, whilst paging-in all the content of a
previous file is comparatively expensive.

        Tony Proctor

"NickHK" <TungCheWah@Invalid.com> wrote in message
news:e#ORakdWEHA.3284@TK2MSFTNGP12.phx.gbl...
> J,
> ~4 minutes for a copy as opposed to ~90 seconds for a create.
>
> NickHK
>
> "J French" <erewhon@nowhere.com> wrote in message
> news:40da9c34.8524506@news.btclick.com...
> > On Thu, 24 Jun 2004 16:53:42 +0800, "NickHK" <TungCheWah@Invalid.com>
> > wrote:
> >
> > >J,
> > >Possibly 100 times that for Duane's initial 20Meg file.
> >
> > It would be 100 times - as it is doing 100 times the work
> > - on the FAT alone
> >
> > Out of interest, and only if you have enough time and disk space,
> > could you /copy/ the large file and see whether that takes the same
> > time as creating it
> > - that is not a totally accurate test, but if copy takes significantly
> > longer than creation, we can be fairly sure that it is not physically
> > writing nulls to disk.
> >
>
>



Relevant Pages

  • Re: LVM write performance
    ... it and finally write it back to the disk. ... noop [deadline] cfq ... Writing 1 GB ... AACRAID* Controller on a Supermicro. ...
    (Debian-User)
  • Re: LVM write performance
    ... When writing to a logical volume directly through the ... you didn't mentioned what disk drive is in use in this test. ... Security Mode feature set ... WRITE_BUFFER command ...
    (Debian-User)
  • Temporary lock-up under heavy write, MegaRAID RAID-5
    ... LSI Logic MegaRAID 320-1, 64MB cache ... when writing very large files. ... disk write access, actually). ... The RAID controller is currently set to "write-through". ...
    (Debian-User)
  • Re: X-CD Roast question again please
    ... Using generic SCSI-3/mmc CD-R driver. ... Disk sub type: Medium Type A, ... ATIP start of lead in: -11318 ... writing 152 KB of pad data. ...
    (RedHat)
  • Re: efficacy of Linux w/o swap
    ... | Caching is supposed to happen, and this does not affect performance. ... that the impact starts to degrade the writing process itself. ... | Writing directly to disk will substantially decrease performance. ... Performance for the first 3/4 of an IDE disk ...
    (comp.os.linux.development.system)