Skip to Main Content

Hardware

Announcement

For appeals, questions and feedback about Oracle Forums, please email oracle-forums-moderators_us@oracle.com. Technical questions should be asked in the appropriate category. Thank you!

Questions about "Vdbench controlled workload maxing out at one million iops" in 5.04.03

Bryce GuintaJul 1 2016 — edited Jul 1 2016

Hi Henk,

In an earlier this year you mentioned that Vdbench controlled workloads maxed out for Vdbench <= 5.04.03

here:

(archived)

So if you set iorate=MAX you have an uncontrolled workload, so this wouldn't apply then?

How about if you are writing to multiple block-devices? For example,

a Vdbench parameter file(for an uncontrolled, multi lun workload such as):

sd=s1,lun=/dev/nvme1n1,align=4096,openflags=o_direct

sd=s2,lun=/dev/nvme0n1,align=4096,openflags=o_direct

wd=wd1,sd=s1,xfersize=4096,seekpct=100

wd=wd2,sd=s2,xfersize=4096,seekpct=100

rd=rd1,wd=wd*,iorate=max,rdpct=100,threads=512,warmup=300,elapsed=300,interval=1

Would have an implicit skew of 50.00 to each lun(from logfile.html):

04:20:21.986 ++Interarrival time: wd=wd1 arrival= 2 rdiorate=999988 target_rate=999988 skew=50.00, wgiorate= 0 wgprio=0 iorate=499994

04:20:21.986 ++Interarrival time: wd=wd2 arrival= 2 rdiorate=999988 target_rate=999988 skew=50.00, wgiorate= 0 wgprio=0 iorate=499994

So does that mean this workload is limited to 1 Million IOPS in vdbench 5.04.03? or 499994?

reqrate is set to 999999 in the flatfile.html. Does this have any meaning for an uncontrolled workload?

Thanks!

Comments
Locked Post
New comments cannot be posted to this locked post.
Post Details
Locked on Jul 29 2016
Added on Jul 1 2016
0 comments
744 views