Yes, thanks Sitsofe. Manual calculation of the size in bytes as input to “rate” helps.
Thanks
+GV
On 4/6/17, 4:17 PM, "Sitsofe Wheeler" <sitsofe@xxxxxxxxx> wrote:
HI,
On 6 April 2017 at 17:46, GV Govindasamy <gv.govindasamy@xxxxxxxxxxx> wrote:
> Thank you, Yes. when I remove “rate_process=poisson”, the IOPS are down to about 85% - 90%.
>
> I checked with rate_iops=100 and rate_iops=1000. In both cases, I got 85%-90% of rate_iops given.
>
> How do I accomplish 100% of IOPS?
> > [db-oltp-w]
> > bssplit=8k/90:16k/10,,
> > size=128G
> > filename=/dev/sdg
> > rw=randread
> > iodepth=8
> > rate_iops=100
It seems to be a quirk of using bsplit with rate_iops. If the
difference between the bsplit sizes is larger then the drop in the
iops achieved will be even higher...
--
Sitsofe | http://sucs.org/~sits/
��.n��������+%������w��{.n�������^n�r������&��z�ޗ�zf���h���~����������_��+v���)ߣ�
