xfs
[Top] [All Lists]

Re: FS Limits (fwd)

To: Seth Mos <knuffie@xxxxxxxxx>
Subject: Re: FS Limits (fwd)
From: Steve Lord <lord@xxxxxxx>
Date: 13 May 2003 11:15:30 -0500
Cc: Oliver Tennert <tennert@xxxxxxxxxxxxxxxxxxxx>, linux-xfs@xxxxxxxxxxx
In-reply-to: <4.3.2.7.2.20030513104210.03f1b0e0@pop.xs4all.nl>
Organization:
References: <4.3.2.7.2.20030513104210.03f1b0e0@pop.xs4all.nl>
Sender: linux-xfs-bounce@xxxxxxxxxxx
On Tue, 2003-05-13 at 03:46, Seth Mos wrote:
> At 09:46 13-5-2003 +0200, Oliver Tennert wrote:
> 
> >I have some questions and am wondering if you could help me any further:
> >
> >it is stated that XFS has a file system size limit of 9 EB. But I could
> >imagine that this could refer to a fs block size of 64k, which is possible
> >with IRIX, but impossible with Linux. Does this mean that with Linux, XFS
> >has effectively a limit of 1/16th of that value, which is something about
> >512 PB?
> 
> AFAIK the limit is 2^63 which equals the 9EB. I believe this to be based on 
> the standard 4K block size.
> 
> We first need a Linux kernel that can scale past the 2TB block device 
> boundary.
> 

Irix can do the insanely large sizes - although there are practical
limitations which come into play before you can get there. We do have
customers with filesystems on the order of 100 Tbytes in size.

On ia32 linux with the LBD patches, you will max out at 16 Tbytes as
this is the address limit of the address_space used to cache the
metadata. Systems with more than 32 bits for the page->index field
can go larger than that.

This does not have anything to do with the fs blocksize used.

Steve

-- 

Steve Lord                                      voice: +1-651-683-3511
Principal Engineer, Filesystem Software         email: lord@xxxxxxx


<Prev in Thread] Current Thread [Next in Thread>