| 1 |
On Wednesday 23 March 2011 08:50:14 Stéphane Guedon wrote: |
| 2 |
> On Wednesday 23 March 2011 08:27:53 Alan McKinnon wrote: |
| 3 |
> > But considering that the thread is all about "what is the best |
| 4 |
> > filesystem?", that too is to be expected. The very title belies a lack |
| 5 |
> > of understanding - the best filesystem for you is the one you have |
| 6 |
> > tested and found best suits your needs. |
| 7 |
> |
| 8 |
> A filesystem looks like quite hard to test (as a kernel, as an hardware... |
| 9 |
> much more complicated than a software you only need to install) : you need |
| 10 |
> a specific machine to test on it. Which tests/operation to perform ? |
| 11 |
> |
| 12 |
> Before launching tests, maybe asking advices to others to have their |
| 13 |
> experiences would be a great idea ! |
| 14 |
> |
| 15 |
> But I would like really to know : can you give a way to test such things ? |
| 16 |
> (hardware ... quite hard : need to buy before testing, kernel, FS). |
| 17 |
> |
| 18 |
> Best regards |
| 19 |
|
| 20 |
no, fs testing is easy. You know what the machine is going to do - so let it |
| 21 |
do it and measure the time it needs. Easy. |
| 22 |
|
| 23 |
That way I found that reiser4+lzo is the best one *for me* and xfs the worst. |
| 24 |
|
| 25 |
But I am sure a lot of people have scenarios where xfs is the best. Or ext4. |
| 26 |
|
| 27 |
And if you don't care about barriers, jfs might be a good choice. |
| 28 |
|
| 29 |
It depends on the stuff you want to do and what do you expect from a file |
| 30 |
system. |
| 31 |
|
| 32 |
Btw, when doing a copy or move test to prime the fs - copy from the same type |
| 33 |
of filesystem or the numbers are skewed. |
Archives