Hi Henk,
I want to generate a load pattern that created of 2k of data and then skipping 2k during sequential writes.
my profile is :
messagescan=no
hd=LGDRM2758,system=LGDRM2758,vdbench=/usr/bin,shell=ssh,user=root,jvms=10
hd=LGDRM2757,system=LGDRM2757,vdbench=/usr/bin,shell=ssh,user=root,jvms=10
sd=default,size=18927616M,openflags=o_direct
sd=sd1,host=LGDRM2757,lun=/dev/mapper/3514f0c5224e00009
wd=wd_oltp2,sd=*,xfersize=(2k,100),seekpct=0,stride=(4,4),rdpct=0
rd=OLTP2,wd=*,iorate=max,interval=1,threads=64
I used stride min = 4 and max = 4 since lba =512B (4 * 512B = 2K)
I've notice by using dd (read data from device) that no skipping occurred
Please advise,
Thanks!