Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On 2005-01-14T11:53:41, Philipp Reisner <philipp.reisner at linbit.com> wrote:
> > How much overhead goes into keeping the active log etc uptodate? Is the
> > performance disk IO-, network IO- or CPU-limitted?
> Ahhh, right. I had the AL at its default setting, so there were updates
> to the AL, because I wrote 3G in my tests...
Ah, I see.
> Now the "al:" value stays constant during a test....
>
> 43.25 MB/sec (3221225472 B / 01:11.033546)
> 44.91 MB/sec (3221225472 B / 01:08.403583)
> 44.79 MB/sec (3221225472 B / 01:08.585322)
>
> Raw write performance to the LVs:
> 59.69 MB/sec (3221225472 B / 00:51.468961)
> 60.54 MB/sec (3221225472 B / 00:50.743192)
> 60.71 MB/sec (3221225472 B / 00:50.603030)
>
> So, when the working set stays constant, we now we reach 75% of the raw
> LVs performance ... with these machines.
Hm. That's still a significant loss. Do you know where the speed is
lost?
I need to build new kernels today anyway, I can test on some lower-end
nodes today and see what happens.
Sincerely,
Lars Marowsky-Brée <lmb at suse.de>
--
High Availability & Clustering
SUSE Labs, Research and Development
SUSE LINUX Products GmbH - A Novell Business