Max usable disk space

[Mike McCarty]

James Wilkinson wrote:
> Mark Haney wrote:
> 
>>Does anyone know the deal 
>>with FDISK?  Are there plans to fix that little problem?  I mean pretty 
>>soon 2TB is gonna be a drop in the bucket for most filesystems, and 
>>considering ext3 handles 64TB or more, this seems a horrible limitation.
> 
> 
> Dave Jones referred to it up-thread. For example:
> http://www.redhat.com/docs/manuals/enterprise/RHEL-4-Manual/release-notes/as-zseries/RELEASE-NOTES-U1-s390x-en.html
> says:
>     · The commonly-used MS-DOS partition table format can not be used on
>       devices larger than 2 TB. For devices larger than 2 TB, the GPT
>       partition table format must be used. The parted utility must be
>       used for the creation and management of GPT partitions. To create
>       a GPT partition, use the parted command mklabel gpt.
> 
> Don't worry about the "zseries" or the "s390" in the URL: that's just
> the system that's most likely to hit this issue. It's talking about the
> same partition table format.
> 
> It's not a bug in fdisk: it does exactly what it's supposed to do --
> handle MS-DOS style partition tables. It's those tables that can't
> handle devices larger than 2 TB -- there just isn't enough space in the
> fields. It's a harder limit than 640K was for real-mode DOS, or 2 GB was
> for FAT16. Change the partition table format so it can handle larger
> devices, and *it's not DOS partition table format any more*, and
> operating systems that don't know about the new format can't use the
> disk. At all.

I don't agree. It is certainly very similar to the differences between
FAT12 and FAT16. A FAT16 FAT is definitely not FAT12 at all. But
that doesn't mean that FAT16 is a whole 'nother ball of wax.

> And if you're going to go for something other than traditional MS-DOS
> partitions, you might as well go for something with a reasonable -- er,
> *any* design, that doesn't have the archaic "4 primary, one of which
> may be extended, and umpteen logical" limitations. Like GPT.

I'm not familiar with GPT. However, to use larger discs, something must
certainly be done. It also certainly makes sense that, if the old
format cannot be "jiggered" up, one reconsider the circumstance from
scratch.

> Note that none of this refers to the format of filesystems within those
> partitions (ext3, reiser, FAT32, etc). It refers to the way the
> partitions themselves are defined and stored on disk. The partition
> table chops up the disk into partitions, and then you put "normal"
> filesystems in the partitions. [1]

Well, mostly, if the first sector on the disc ends in AA55 the BIOS
will almost certainly load it, and one can put any boot manager on
there one wants.

> 
> James.
> 
> [1] Or not. Swap partitions and raw device access are beyond the scope
> of this e-mail...

I'm not sure that swap partitions make sense any more. I've seen
data that indicates that, at least with ext3, using a swap file
is no slower than using a partition.

Mike
-- 
p="p=%c%s%c;main(){printf(p,34,p,34);}";main(){printf(p,34,p,34);}
This message made from 100% recycled bits.
You have found the bank of Larn.
I can explain it for you, but I can't understand it for you.
I speak only for myself, and I am unanimous in that!