[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: OT: Comparison of filesystems



Curt Howland wrote:

[snip]

My personal opinion is that anything "up to date" (as opposed to, say, FAT12) will provide decent service for a desktop machine. I would add

Seconded.

journaling, which is why I also use ext3, but with the caveat that ext3 is just an add-on to ext2. Performance demonstrates this.

I use ext3, but have disabled simply coming up on the journal after
powerfail/crash.

If I were to decide on something more than just ext3, I think I would go with ReiserFS "just because". It seems quite flexible, and while it does require a periodic "defragmentation", well, such is life. Ext2/3 doesn't require defragmentation, but they are also not as optimized in their performance. There are always trade-offs.

Reiser is somewhat faster than ext3, but has much less error recovery
toolset. It is also somewhat better at actual disc usage for many
small files. OTOH, when large files are involved, it seems to be
somewhat slower than ext3.

IMO, "journaling" provides *no* advantage, and some significant
disadvantages for small file systems. For small discs (like
a few 100 GB or so) a full fsck takes very little time anyway, and is
much better than using a journal. Journaling is sometimes touted
as being "atomic action" resulting in an "incorruptible" file
system. This is just untrue, and the journal, if corrupt, can
result in permanent damage when used. For large file systems,
there can be a significant advantage in speed of recovery from
power fail. This may justify the lesser confidence that using
the journal rather than fsck introduces.

When a powerfail occurs, one does not know what has happened
to the discs. Writes may take place anywhere on the surface.
Using a journal on next reboot may result in the disc being
declared "clean" after applying the journal, but it may actually
be in a very sad state. Using fsck is much better at detecting
errors.

If you need atomic disc access, this can be done, but it
requires more than a journal.

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!



Reply to: