CentOS 5.4 vs. OpenSuSE 11.2 vs. Ubuntu 9.10 Benchmarks

[kraftman]

openSUSE uses barriers enabled, file extents enabled, delayed allocation enabled, mballoc enabled, ordered data mode. internal journal by default on ext4.

Thanks Deanjo. It seems (K)ubuntu also uses ordered data mode and I'll test writeback to see if it will make a difference in SQLite.

[deanjo]

Thanks Deanjo. It seems (K)ubuntu also uses ordered data mode and I'll test writeback to see if it will make a difference in SQLite.

Barriers enabled does seem to have quite an effect with SQLite. You might want to check that as well. IIRC though (K)ubuntu does not have it enabled where openSUSE does.

[energyman]

well, extX devs only care about good benchmark results. So the 'defaults' are 'fast but very insecure'. AKA completly moronic.

A dsitro using 'just defaults' might look better on paper, but puts your data at risk. A completly idiotic behaviour.

Seriously, using extX (2,3,4) with default ops says 'I don't care about data. Just pretty numbers'.

Another reason, why ext4 should have never become stable in the first place.

[EthraZa]

Phoronix: CentOS 5.4 vs. OpenSuSE 11.2 vs. Ubuntu 9.10 Benchmarks

With the release of CentOS 5.4 last month to bring this community enterprise operating system on par with Red Hat Enterprise Linux 5.4, we decided it was a good time to see how the server / workstation performance between this new CentOS release compares to that of Ubuntu 9.10, which was released last week, and also how it performs up against the release candidate of OpenSuSE 11.2. In this article are these benchmarks.

http://www.phoronix.com/vr.php?view=14332

Where is the Mandriva Corporate Server in this tests?
It wold probably win the benchmark!

[deanjo]

Where is the Mandriva Corporate Server in this tests?
It wold probably win the benchmark!

I doubt they have anymore server oriented optimizations when compared to CentOS.

[RealNC]

well, extX devs only care about good benchmark results. So the 'defaults' are 'fast but very insecure'. AKA completly moronic.

A dsitro using 'just defaults' might look better on paper, but puts your data at risk. A completly idiotic behaviour.

Seriously, using extX (2,3,4) with default ops says 'I don't care about data. Just pretty numbers'.

Another reason, why ext4 should have never become stable in the first place.

Hmm? I never changed the defaults on Gentoo and I get "barrier=1,data=ordered".

[EthraZa]

At least in the desktop, Mandriva is one of the fastest. Since the Server version is based on the Desktop version plus server specifics, I imagine that the Mandriva Corporate Server is probably one of the fastest if not The fastest.
Any way we cold be sure if Phoronix do the same tests with it and add the results to the comparison.

[deanjo]

At least in the desktop, Mandriva is one of the fastest. Since the Server version is based on the Desktop version plus server specifics, I imagine that the Mandriva Corporate Server is probably one of the fastest if not The fastest.
Any way we cold be sure if Phoronix do the same tests with it and add the results to the comparison.

Not sure where you get the idea of Mandriva being "one of the fastest". Previous versions actually show it to be one of the slowest on a fairly consistent basis.

Do not confuse the loading of a gui with over all system speed.

http://www.phoronix.com/scan.php?page=article&item=distro_four_way&num=1

[deanjo]

Can you please list which kernel you were using for the server/workstation benchmarks for openSUSE 11.2?

There are two flavors,

-desktop (which is optimized and configured for desktop use)
-default (which optimized and configured for workstation/server use)

By default openSUSE 11.2 will install the desktop kernel so unless it was changed from the default install you would have had the 2.6.31-desktop kernel installed when running your tests.

PING! Inquiring minds want to know....

[kraftman]

Barriers enabled does seem to have quite an effect with SQLite. You might want to check that as well. IIRC though (K)ubuntu does not have it enabled where openSUSE does.

I'd want to, but SQLite benchmark doesn't work properly here. It looks like it's caught in some loop and it doesn't want to finish. I ran it using Ext4 and default settings. Pgbench works fine.