The final dedup ratio doesn't match the target dedupratio in vdbench configuration file

Question

Hi,
I would like to fill-up the volumes with predefined dedup ratio using vdbench dedupratio and dedupunit values introduced in vdbench 5.03
I'm using 2 centos servers running vdbench50402 (also tried vdbench50403rc1). Below you can see the sample configuration file.
With all the dedup values i receive constantly 1.1 dedup ratio on the storage array. What are the possible reasons for incorrect final dedup ratio  ??
Thanks,
Alex I.
#********Vdbench configuration file*************************************
dedupratio=5
dedupunit=8k
hd=default,vdbench=/vdbench50402,user=root,shell=vdbench
hd=hostwg1,jvms=20,system=host-wg-1
hd=hostwg2,jvms=20,system=host-wg-2
#Define the test device size as "size=xxx"
sd=default,size=2136G,openflags=o_direct
sd=sd1hostwg1,lun=/dev/mapper/123456789010007a,host=hostwg1
sd=sd2hostwg1,lun=/dev/mapper/123456789010007b,host=hostwg1
sd=sd3hostwg1,lun=/dev/mapper/123456789010007c,host=hostwg1
sd=sd4hostwg1,lun=/dev/mapper/123456789010007d,host=hostwg1
sd=sd5hostwg1,lun=/dev/mapper/123456789010007e,host=hostwg1
sd=sd6hostwg1,lun=/dev/mapper/123456789010007f,host=hostwg1
sd=sd7hostwg1,lun=/dev/mapper/1234567890100080,host=hostwg1
sd=sd8hostwg1,lun=/dev/mapper/1234567890100081,host=hostwg1
sd=sd1hostwg2,lun=/dev/mapper/1234567890100082,host=hostwg2
sd=sd2hostwg2,lun=/dev/mapper/1234567890100083,host=hostwg2
sd=sd3hostwg2,lun=/dev/mapper/1234567890100084,host=hostwg2
sd=sd4hostwg2,lun=/dev/mapper/1234567890100085,host=hostwg2
sd=sd5hostwg2,lun=/dev/mapper/1234567890100086,host=hostwg2
sd=sd6hostwg2,lun=/dev/mapper/1234567890100087,host=hostwg2
sd=sd7hostwg2,lun=/dev/mapper/1234567890100088,host=hostwg2
sd=sd8hostwg2,lun=/dev/mapper/1234567890100089,host=hostwg2
wd=wd_fill,sd=sd*,xfersize=256k,rdpct=0
rd=default
rd=fill_pass1,wd=wd_fill,iorate=max,interval=1,elapsed=172800
#********Vdbench configuration file*************************************

Answer 1

'dedupunit=' is the key here.
The Vdbench dedup logic is based around "storage recognizes duplicate data blocks in chunks of 8k".
If your storage has a different 'dedup unit' Vdbench won't create the proper data patterns.
An other issue may be the FILLING and reporting of your luns: if you have a 10tb lun but Vdbench only writes 1tb, then your dedup will only include 1TB worth of Vdbench written data patterns.
Note: May I suggest in above parameter file to start using 'seekpct=eof'? Vdbench will then stop after the last block on all SDs has been written instead of your current 48 hours, which could result in rewriting the same blocks over and over again, because with seekpct=0 Vdbench just starts at the beginning again when it reaches the end.
Of course, if 48 hours is not enough elapsed time Vdbench will terminate BEFORE the last block has been written.
Hope this helps.
Henk.