Non-intuitive gains in performance

[Russ Allbery]

Bill Davidsen <davidsen at tmr.com> writes:

> I'm running some servers with an o/s which has a 2GB file limit. I
> thought I would get better performance if I used the cycbuffs
> sequentially, since the storage is on a RAID controller with a stripe
> size too small to work well. The idea was to at least thrash in one part
> of the disk:-(

> So I set the metacycbuff SEQUENTIAL as advertised, and the artwrite time
> didn't go down a bit. However, the hiswrite time dropped by some 40%! 
> Does anyone have a thought on this? It's nice that I really helped the
> performance, but I lack any idea why it works this way.

Is history on the same drives as the storage?  What's the disk
configuration?

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>