onstat -g iof sanity check

onstat -g iof output ... The "avg. time" field (under each chunk) is in milliseconds, correct? So a value of 0.0860 = 86ms? I'm seeing some shocking numbers but want to confirm first. IO/s vs. service times are inverted, which would make sense. What are some baseline metrics for service times and IO/S on SSD's? I remember all sorts of numbers from over the years but never for SSDs. I'm seeing as low as 11 IO/S and as high as 86ms service times for kaio reads. Art will enjoy this ... we're still on RAID 5 and the chance of convincing them of it's "bad choice" is slim. But I'm hoping Art will chime in here with respect to the iof output. Thanks - Mark Scranton The Mark Scranton Group mark@markscranton.com

Don't know why I haven't seen Mark's post in my email. Anyway, I don't have any hard numbers for SSD drives either Mark. For spindles I'd like to see service times around 10ms (0.010) or better with 25ms being my PAIN level. I would expect that SSDs would be an order of magnitude faster so single digits for sure. Art

If you are using a newer version (late 11.7 or any 12) of Informix,&= nbsp; I would start using onstat -g ioh instead of onstat -g iof.&nbs= p; The information is about the same, but with onstat -g ioh you have= 60 one minute averages for each chunk. With onstat -g iof you = have very long running averages and it is very hard to catch spikes.<= br> John F. Miller III STSM, Lead Architect [1]miller3@us.ibm.com 503-747-1366 IB= M Informix Dynamic Server (IDS) [2]-----ids-bounce= s@iiug.org wrote: ----- >To: ids@iiug.org >From: "MARK SCRANTON" >Sent by: [3]= ids-bounces@iiug.org >Date: 12/02/2015 06:35PM >Subject: on= stat -g iof sanity check [36139] > >onstat -g iof output ... <= br>> >The "avg. time" field (under each chunk) is in milliseconds,= correct? >So a >value of 0.0860 =3D 86ms? I'm seeing some sho= cking numbers but want to >confirm >first. IO/s vs. service ti= mes are inverted, which would make sense. > >What are some bas= eline metrics for service times and IO/S on SSD's? I >remember &g= t;all sorts of numbers from over the years but never for SSDs. I'm >s= eeing as low >as 11 IO/S and as high as 86ms service times for kaio = reads. > >Art will enjoy this ... we're still on RAID 5 and th= e chance of >convincing >them of it's "bad choice" is slim. Bu= t I'm hoping Art will chime in >here with >respect to the iof = output. > >Thanks - >Mark Scranton >The Mark Scr= anton Group >[4]mark@markscranton.com > > >*******************= ************************************************** >********** &g= t; Forum Note: Use "Reply" to post a response in the discussion forum. = > > > References 1. =3D"mailto:miller3@us.ibm.com" 2. 3D"mailto:-----ids-bounces@iiug.org" 3. 3D"mailto:ids-bounces@iiug.org" 4. 3D"mailto:mark@markscranton=

I have done some SSD testing recently within a customer test environment and we are seeing sub-millisecond response times. It is an easy but not necessarily cheap way to see a significant performance increase in large systems that have IO issues Cheers Paul Paul Watson Oninit www.oninit.com +1 913 387 7529 Oninit® is a Registered Trademark of Oninit LLC > On Dec 2, 2015, at 17:43, ART KAGEL <art.kagel@gmail.com> wrote: > > Don't know why I haven't seen Mark's post in my email. Anyway, I don't have > any hard numbers for SSD drives either Mark. For spindles I'd like to see > service times around 10ms (0.010) or better with 25ms being my PAIN level. I > would expect that SSDs would be an order of magnitude faster so single digits > for sure. > > Art > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum.

BTW, onstat -g ioh is still not documented as of 12.10.xC5, not in the Administrator's Reference manual nor even in the onstat usage output. Art Art S. Kagel, President and Principal Consultant ASK Database Management www.askdbmgt.com Blog: http://informix-myview.blogspot.com/ Disclaimer: Please keep in mind that my own opinions are my own opinions and do not reflect on the IIUG, nor any other organization with which I am associated either explicitly, implicitly, or by inference. Neither do those opinions reflect those of other individuals affiliated with any entity with which I am affiliated nor those of the entities themselves. On Thu, Dec 3, 2015 at 11:33 AM, JACQUES RENAUT <jrenaut@us.ibm.com> wrote: > Original post: > > John what version of 11.70 did the onstat -g ioh come in?? > > Bruce Simms > Data Base Services > TALX, Provider of Equifax Workforce Solutions > 2330 Ball Drive > St. Louis, MO 63146 > > Phone (314) 214-7703 > Cell (314) 496-6883 > FAX (314) 983-3238 > Bruce.Simms@Equifax.com > IBM Champions > > Reply: > > I'm not seeing the -g ioh in the 11.70 code at all, not even the last > 11.70.FC8W1 release. It does show up immediately in 12.10.xC1 though. > > Jacques Renaut > IBM Informix Advanced Support > APD Team > > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > > --001a113fe51e551eda0526011b83