hi Phil, thanks for the url... I have read those url 3 times before this trussing :) do you think that statement comes from oracle kernel guy? or because of open flag for O_DSYNC 28355: open("/data01/zzztest1.dbf", O_RDWR|O_NDELAY|O_DSYNC|O_NONBLOCK) = 10 that's why asynch io didn't work? On 3/22/12, Phillip Jones <phil@xxxxxxxxxx> wrote: > Hi, > > You may find this archive post/thread of interest: > //www.freelists.org/post/oracle-l/Async-and-Direct-IO-On-Solaris-using-VXFS,5 > > Cheers, > > Phil > > On Thu, Mar 22, 2012 at 6:04 AM, Ujang Jaenudin > <ujang.jaenudin@xxxxxxxxx>wrote: > >> dear all, >> I have a puzzle things for this environment. >> >> I tested create table as select/CTAS (the original table is on QIO file) >> it looks no kaio AIOREAD, but uses pread (synchronous IO). >> >> play with db parameter: >> filesystemio_options >> disk_asynch_io >> >> has no effects. >> >> anyone had same experiences? >> do you think oracle has some issues on this veritas QIO to use kaio >> AIOREAD >> <//www.freelists.org/webpage/oracle-l> >> >> >> > -- regards, ujang jaenudin databases independent consultant http://ora62.wordpress.com jakarta - Indonesia -- //www.freelists.org/webpage/oracle-l