[linux-lvm] disk capacity discrepancy: is it 1000 versus 1024?

[linux-lvm] disk capacity discrepancy: is it 1000 versus 1024?

[Maurice Volaski]

I originally had a hardware RAID with fourteen 400 GB drives of 
usable capacity (Hitachi 7K400) . That's 5600 GB altogether. That's 
what both fdisk and df report (filesystem was ext3).

When I instead created an LVM2 physical volume on the device, I get 
just 5.09 TB.

Is LVM eating several hundreds gigabytes of space or is it the case 
that LVM uses terminology in a strict multiply by 1024 fashion versus 
everything else, which is multiplying by 1000. In that case, I guess 
the drives aren't really 400 GBs. :-(

However, it appears that parameters passed to LVM commands are 
interpreted using x1000, not x1024.

other keywords: gigabinarybytes, gibibytes, Gi, terabinarybytes, tebibytes, Ti.
-- 

Maurice Volaski, mvolaski@aecom.yu.edu
Computing Support, Rose F. Kennedy Center
Albert Einstein College of Medicine of Yeshiva University

Re: [linux-lvm] disk capacity discrepancy: is it 1000 versus 1024?

[Greg Freemyer]

On 5/14/05, Maurice Volaski wrote:
> I originally had a hardware RAID with fourteen 400 GB drives of
> usable capacity (Hitachi 7K400) . That's 5600 GB altogether. That's
> what both fdisk and df report (filesystem was ext3).
> 
> When I instead created an LVM2 physical volume on the device, I get
> just 5.09 TB.
> 
> Is LVM eating several hundreds gigabytes of space or is it the case
> that LVM uses terminology in a strict multiply by 1024 fashion versus
> everything else, which is multiplying by 1000. In that case, I guess
> the drives aren't really 400 GBs. :-(
> 
> However, it appears that parameters passed to LVM commands are
> interpreted using x1000, not x1024.
> 
> other keywords: gigabinarybytes, gibibytes, Gi, terabinarybytes, tebibytes, Ti.
> --
> 
> Maurice Volaski, mvolaski@aecom.yu.edu
> Computing Support, Rose F. Kennedy Center
> Albert Einstein College of Medicine of Yeshiva University

As Maurice implies NIST is making an effort (however small) to
standardize GB as 1000 * 1000 * 1000 bytes.

See http://physics.nist.gov/cuu/Units/binary.html

So per NIST:
GB = Gigabyte = 1,000 * 1,000 * 1,000 Bytes
GiB = Gibibyte  = 1,024 * 1,024 * 1,024 Bytes

Or for Maurice TB vs. TiB

BTW, the difference between TB and TiB is almost 10% so this has a
very user observable effect and would explain Maurice's discrepancy.

So if I understand Maurice's question, it becomes:
   "Is LVM reporting TiB units with a TB designation?"

Greg
-- 
Greg Freemyer
The Norcross Group
Forensics for the 21st Century

Re: [linux-lvm] disk capacity discrepancy: is it 1000 versus 1024?

[Greg Freemyer]

On 5/14/05, Greg Freemyer wrote:
> On 5/14/05, Maurice Volaski wrote:
> > I originally had a hardware RAID with fourteen 400 GB drives of
> > usable capacity (Hitachi 7K400) . That's 5600 GB altogether. That's
> > what both fdisk and df report (filesystem was ext3).
> >
> > When I instead created an LVM2 physical volume on the device, I get
> > just 5.09 TB.
> >
> > Is LVM eating several hundreds gigabytes of space or is it the case
> > that LVM uses terminology in a strict multiply by 1024 fashion versus
> > everything else, which is multiplying by 1000. In that case, I guess
> > the drives aren't really 400 GBs. :-(
> >
> > However, it appears that parameters passed to LVM commands are
> > interpreted using x1000, not x1024.
> >
> > other keywords: gigabinarybytes, gibibytes, Gi, terabinarybytes, tebibytes, Ti.
> > --
> >
> > Maurice Volaski, mvolaski@aecom.yu.edu
> > Computing Support, Rose F. Kennedy Center
> > Albert Einstein College of Medicine of Yeshiva University
> 
> As Maurice implies NIST is making an effort (however small) to
> standardize GB as 1000 * 1000 * 1000 bytes.
> 
> See http://physics.nist.gov/cuu/Units/binary.html
> 
> So per NIST:
> GB = Gigabyte = 1,000 * 1,000 * 1,000 Bytes
> GiB = Gibibyte  = 1,024 * 1,024 * 1,024 Bytes
> 
> Or for Maurice TB vs. TiB
> 
> BTW, the difference between TB and TiB is almost 10% so this has a
> very user observable effect and would explain Maurice's discrepancy.
> 
> So if I understand Maurice's question, it becomes:
>    "Is LVM reporting TiB units with a TB designation?"
> 
> Greg
> --
> Greg Freemyer
> The Norcross Group
> Forensics for the 21st Century
> 
Maurice,

I forgot to add that disk drive manufactures have been selling disks
by the GB for years, not the GiB.  Thus your 400 GB disk drive is most
likely about 373 GiB.  (I used a calculator).

One way to tell for sure is to get the number of sectors and multiply
it out.  Sectors are typically 512 bytes.  You can typically see the
sector count or the paper label of newer drives, or with standard ide
disks in /proc/ide/hdx/capacity.

Greg

-- 
Greg Freemyer
The Norcross Group
Forensics for the 21st Century

RE: [linux-lvm] disk capacity discrepancy: is it 1000 versus 1024?

[Christian Hack]

> -----Original Message-----
> From: linux-lvm-bounces@redhat.com 
> [mailto:linux-lvm-bounces@redhat.com] On Behalf Of Greg Freemyer
> Sent: Sunday, 15 May 2005 7:42 AM
> To: LVM general discussion and development
> Subject: Re: [linux-lvm] disk capacity discrepancy: is it 
> 1000 versus 1024?
> 
> 
> On 5/14/05, Greg Freemyer wrote:
> > On 5/14/05, Maurice Volaski wrote:
> > > I originally had a hardware RAID with fourteen 400 GB drives of
> > > usable capacity (Hitachi 7K400) . That's 5600 GB 
> altogether. That's
> > > what both fdisk and df report (filesystem was ext3).
> > >
> > > When I instead created an LVM2 physical volume on the 
> device, I get
> > > just 5.09 TB.
> > >
> > > Is LVM eating several hundreds gigabytes of space or is 
> it the case
> > > that LVM uses terminology in a strict multiply by 1024 
> fashion versus
> > > everything else, which is multiplying by 1000. In that 
> case, I guess
> > > the drives aren't really 400 GBs. :-(
> > >
> > > However, it appears that parameters passed to LVM commands are
> > > interpreted using x1000, not x1024.
> > >
> > > other keywords: gigabinarybytes, gibibytes, Gi, 
> terabinarybytes, tebibytes, Ti.
> > > --
> > >
> > > Maurice Volaski, mvolaski@aecom.yu.edu
> > > Computing Support, Rose F. Kennedy Center
> > > Albert Einstein College of Medicine of Yeshiva University
> > 
> > As Maurice implies NIST is making an effort (however small) to
> > standardize GB as 1000 * 1000 * 1000 bytes.
> > 
> > See http://physics.nist.gov/cuu/Units/binary.html
> > 
> > So per NIST:
> > GB = Gigabyte = 1,000 * 1,000 * 1,000 Bytes
> > GiB = Gibibyte  = 1,024 * 1,024 * 1,024 Bytes
> > 
> > Or for Maurice TB vs. TiB
> > 
> > BTW, the difference between TB and TiB is almost 10% so this has a
> > very user observable effect and would explain Maurice's discrepancy.
> > 
> > So if I understand Maurice's question, it becomes:
> >    "Is LVM reporting TiB units with a TB designation?"
> Maurice,
> 
> I forgot to add that disk drive manufactures have been selling disks
> by the GB for years, not the GiB.  Thus your 400 GB disk drive is most
> likely about 373 GiB.  (I used a calculator).
> 
> One way to tell for sure is to get the number of sectors and multiply
> it out.  Sectors are typically 512 bytes.  You can typically see the
> sector count or the paper label of newer drives, or with standard ide
> disks in /proc/ide/hdx/capacity.
> 

Although Greg has bascially done the maths, it works out exactly right as
5.09TiB

i.e. 14 * 400 * 10^9 = 5600 * 10^9 = 5.6 * 10^12 TB

In TiB

(5.6 * 10 ^12) / (1024 * 1024 * 1024 * 1024) = 5.09 TiB which is exactly
what you are seeing. 

I would guess LVM is simply reporting TiB with the suffix TB since the whole
Gibi, Tibi etc thing is yet to completely take off and not everyone is using
it.

It starts getting a bit rough when you lose 500GB due to 1024 vs 1000.

CH