| 1 |
On Thu, 2005-12-15 at 14:43 +0100, Francesco Riosa wrote: |
| 2 |
> having more than one disk or a lot of memory add very interesting |
| 3 |
> addition, read raid 0 (stripe) or tmpfs for working data that does'nt |
| 4 |
> need a backup fex: $PORTIR, /var/tmp ... |
| 5 |
tmpfs has miserable performance when larger than RAM iirc - you'd need >5G for openoffice :-) |
| 6 |
|
| 7 |
> I've found that preemption with the new standard 250Hz of the kernel is |
| 8 |
> suitable for mostly needs, however no server here has preemption enabled ;-) |
| 9 |
My system still manages to run a DVD at a load of ~8, so from my point of view that is not a problem |
| 10 |
(2Ghz Athlon ... one of the "faster" machines I'd say as many people |
| 11 |
still use ~500Mhz) |
| 12 |
|
| 13 |
What causes more problems are packages that become slow on update - e.g. |
| 14 |
gtk+ 2.8 is _really_ slow (takes a few seconds to redraw apps that took |
| 15 |
<1sec with 2.6 ... :-( ) |
| 16 |
> what is a normal workload ? Define it and creating tests should not be |
| 17 |
> so difficult, then there are apps that can help to profiling, thinking |
| 18 |
> at bootchart, sysproof, memproof, valgrind ... strace |
| 19 |
I guess then you'd have to split into server / desktop / ... |
| 20 |
|
| 21 |
|
| 22 |
> reiserfs is sustainable, at least for 99.999% of uses, last reiserfs bug |
| 23 |
> on very high load (and with degraded raid5) is dated 4 years ago here. |
| 24 |
> However upstream is going to the route of reiser4, much more complex, |
| 25 |
> and much more unstable, latest problems where in 2.6.14, additionally no |
| 26 |
> devs in gentoo are (will?) support it the patch for grub it's still not |
| 27 |
> in place I think. |
| 28 |
reiser4 is "new and untested", I'd keep away from it until it has shown its reliability. |
| 29 |
Also in my (limited) testing it is relatively slow (about the same speed |
| 30 |
as reiser3) |
| 31 |
|
| 32 |
> > Are there any application-specific tweaks (e.g. "use the prefork MPM |
| 33 |
> > with apache2")? What is known to break things, what has usually |
| 34 |
> > beneficial behaviour? Are there any useful benchmarks that show the |
| 35 |
> > performance difference between different settings? |
| 36 |
> is'n there "ab" [1] for apache testing ? |
| 37 |
Yes, but that's apache specific and is quite hard to use correctly. |
| 38 |
(but very nice for slashdotting simulation ;-) ) |
| 39 |
|
| 40 |
Patrick |
| 41 |
-- |
| 42 |
Stand still, and let the rest of the universe move |
Archives