33 MB overhead with ext3
33 MB overhead with ext3
Posted Aug 26, 2004 4:14 UTC (Thu) by jvotaw (subscriber, #3678)Parent article: Looking at reiser4
I thought the ~33 MB used on a fresh ext3 filesystem was mostly because of the journal (default size of 32 MB on that size of filesystem), not pre-allocated inode tables.
Apologies if I'm wrong, or too ignorant to see that these two are the same thing, or ...
-Joel
33 MB overhead with ext3
Posted Aug 26, 2004 7:02 UTC (Thu)
by dmantione (guest, #4640)
[Link] (5 responses)
Does it matter? Reiserfs does use a journal as well. Posted Aug 26, 2004 7:02 UTC (Thu) by dmantione (guest, #4640) [Link] (5 responses)
33 MB overhead with ext3
Posted Aug 26, 2004 12:28 UTC (Thu)
by mtk77 (guest, #6040)
[Link] (2 responses)
As with inodes, so with journals. reiserfs4 doesn't preallocate the journal either. (*Disclaimer* this as I understand it.)Posted Aug 26, 2004 12:28 UTC (Thu) by mtk77 (guest, #6040) [Link] (2 responses)
33 MB overhead with ext3
Posted Aug 26, 2004 14:31 UTC (Thu)
by erich (guest, #7127)
[Link] (1 responses)
I do not understand why preallocating the journal or inode tables is considered a bad thing.Posted Aug 26, 2004 14:31 UTC (Thu) by erich (guest, #7127) [Link] (1 responses)
In the rare cases where one wants to store single big files on a hard disc, neither ext3 or reiserfs with default options is the best choice.
If the journal is allocated when creating the filesystem it probably is placed at the beginning which usually is the faster area of the disc, isn't it? Also it is contiguous, which should increase performance, too.
(why is using a swap partition better than using a swap file? similar reasons)
But i'm not an expert at all. I just dislike things that are taken for better without giving reasons to do so.
swapfiles
Posted Aug 27, 2004 14:59 UTC (Fri)
by Luyseyal (guest, #15693)
[Link]
Actually, I read awhile back that swap files are now just as fast as swap partitions due to some VFS magic.Posted Aug 27, 2004 14:59 UTC (Fri) by Luyseyal (guest, #15693) [Link]
-l
33 MB overhead with ext3
Posted Aug 26, 2004 15:11 UTC (Thu)
by meuh (guest, #22042)
[Link] (1 responses)
Reiserfs is not a journalised filesystem.Posted Aug 26, 2004 15:11 UTC (Thu) by meuh (guest, #22042) [Link] (1 responses)
It's an atomic one, so there's no need for a journal.
Or I don't understand Hans posts :)
33 MB overhead with ext3
Posted Oct 22, 2004 15:43 UTC (Fri)
by pont (guest, #25575)
[Link]
>Reiserfs is not a journalised filesystem.Posted Oct 22, 2004 15:43 UTC (Fri) by pont (guest, #25575) [Link]
>It's an atomic one, so there's no need for a journal.
>
>Or I don't understand Hans posts :)
An atomic filesystem means a transaction is completed or not completed, not half completed 3/4 completed, etc. etc, To make a filesystem atomic we use logs meta-data logging as in Reiser3 or wandering logs as in Reiser4.