| 1 |
On Fri, 2009-01-30 at 12:49 -0600, Harry Putnam wrote: |
| 2 |
> I didn't want to derail the existing thread discussing ext4 with this |
| 3 |
> angle ... I'm guessing there may be comments that will not be helpful |
| 4 |
> to that OP. |
| 5 |
> |
| 6 |
> I'm wondering what people running ext4 are seeing in practice that |
| 7 |
> makes it better than ext3 or reiserfs? Is it safer journalling? Faster |
| 8 |
> read/write? ... |
| 9 |
> |
| 10 |
> I've thought about switching over too... especially every time I |
| 11 |
> `rm -rf' something big and it seems to take way longer than I'd like. |
| 12 |
> |
| 13 |
> (I run all reiserfs except ext2 for /boot) |
| 14 |
|
| 15 |
Well it's new and new is always interesting (in good ways and bad ;). |
| 16 |
|
| 17 |
Large writes/deletes will be faster. If you don't do (a lot of) |
| 18 |
writes/deletes of large files then you won't notice (as much). Extents, |
| 19 |
better allocation/deallocation methods, and other added logic further |
| 20 |
makes improvements on files (esp large files). It will eventually |
| 21 |
support much larger filesystems and subsecond timestamps for those with |
| 22 |
the need. |
| 23 |
|
| 24 |
Depending on your usage you might see significant improvements or hardly |
| 25 |
any at all. Best way to know for sure is to try it out. Note however |
| 26 |
that on ext4 journal checksums are *on* by default (and off on ext3 |
| 27 |
iirc). So when you are comparing performance you should make that value |
| 28 |
the same for both for a fair comparison. |
Archives