Rayees wrote:
> Hi Paul
>
> The information is really great.
>
> I understood upto certain extent, I downloaded PartinNT and say the disk
> geometry.
>
> I have query here. I have 60GB HDD and I have two partition. C: is Primary
> and D:\ is extended partition.
>
> When I run Winmsd --> Storage --> Disk. I saw Partition starting Offset
> information for Disk 0 partition 0 is 32256 bytes where as Disk 0 partion 1
> is 27,085,847,040 bytes.
>
> If my objective is to change the partition starting offset from 32256 to
> 32768 (as per SAP), and I have constraints in doing that, on what basis Disk
> 0 Partition 1 value is so high.
>
> I'm not ver clear, since I lack basic knowledge in this subject. Will be
> greatful, if you can guide me please.
>
> Regards
> Rayees
>
OK, I tried a little experiment.
I installed my copy of Partition Magic 7 (PowerQuest version) and tested
on a brand new disk. The geometry of my disk is 9729 cylinders, 255 heads, 63 sectors
(an 80GB disk).
If I use Partition Magic, it allows a partition to start at sector 63 (not 64).
If I use the "Move" option, to reposition the partition, the partition moves
in increments of 7.8MB!. Moving the partition by single sectors is not allowed!
The partition can start at sector 63, or at sector 63 + 255*63 etc. The last
value is "sector 63 plus one cylinder" of the fake geometry. I am not allowed to
add just one sector.
I looked up the interface of the RedHat Linux partitioning tool. You can
see here, that the partition offset field in the tool, is in units of
cylinders, and not sectors. Thus, you cannot do it in Linux either. The
Linux tool behaves the same way that PartitionMagic does.
http://www.redhat.com/docs/manuals/linux/RHL-9-Manual/install-guide/s1-diskpartitioning.html
The disk no longer has a geometry. The 9729/266/63 value, satisfies the
traditional interface, until the operating system is running. Once running,
as far as I know, the disk and OS talk "LBA" or logical block addresses,
and those are a simple sector number. (I even looked up in the ATAPI spec,
and while "cylinder" is defined in the terminology section, it does not
appear in the body of the specification itself. So CHS is obsolete.)
http://www.t13.org/FTPSite/Default.aspx (get the userid/password here first)
ftp://ftp.t13.org/docs2002/d1532v1r1a-ATA-ATAPI-7.pdf (then check this document)
ftp://ftp.t13.org/docs2002/d1532v2r1a-ATA-ATAPI-7.pdf
For a primer on disk geometry, try this storagereview article.
http://www.storagereview.com/guide2000/ref/hdd/bios/sizeOlder.html
Also, for another experiment, download HDTach, and run the "long bench".
Click "long bench" and the click "Run Test". The resulting stair step
curve, shows how the actual geometry inside the disk is "zoned" and
is not of a constant density. Thus, the SAP idea, of aligning the disk,
is just foolish. If each zone has a different number of sectors, then
no performance optimization by means of sector offset, is feasable.
http://www.simplisoftware.com/Public/index.php?request=HdTach
Summary:
As near as I can tell, the SAP advice you've been given is *wrong*.
Hope that helps,
Paul