Re: sysptprof questoin ✓ Answered: green (solid confidence) — David explained why isreads increases during a sequential scan (ISAM/C-ISAM layer function calls, not index-only reads) and Mark confirmed it resolved his confusion. Note: the original question message itself is only preserved quoted inside the reply (a threading artifact), not as its own record.

Advisory only.

isreads is ISAM () reads i.e. read calls to functions in the lower (storage) layer of the database engine. This counts internal function calls, not i/o's. The associated i/o's can be from buffers or storage. As per Madison Pruet - "also a single isam read/write can hit multiple buffers... " http://members.iiug.org/forums/ids/index.cgi/read/29195 ISAM comes from the history of Informix - C-ISAM https://en.wikipedia.org/wiki/IBM_Informix_C-ISAM Regards, David. > On 17 February 2017 at 19:58 MARK JALKIEWICZ <mark.jalkiewicz@verizon.net> > wrote: > > > IDS 1150.FC8 > > I am debugging a "hot table contention issue" by extracting info from the > sysptprof table. To test this script I kicked off a session that forced > repeated sequential scans on the table ( "select * from table ") and the > output from sysptprof is what I expected ; but one thing that seems out of > order is that isreads is increasing proportionally with bufreads (about 50% of > > bufreads); an sqexplain does confirm a seq scan. There are two indexes on this > > table. I do clear the metrics (onstat -z) before starting my monitoring. > > Anyone know why isread would increase ? I assumed isreads are soley reads on > index fragments. > > Mark > > > ******************************************************************************* > > Forum Note: Use "Reply" to post a response in the discussion forum. >

Thanks Dave.. that clears that up. Mark