tag:blogger.com,1999:blog-7003976656201910397.post1159311670178888654..comments2024-03-12T09:59:57.644+01:00Comments on External Table: Oracle Events' Latency Visualization and Heat Maps in SQL*plusLuca Canalihttp://www.blogger.com/profile/06252662329568134677noreply@blogger.comBlogger11125tag:blogger.com,1999:blog-7003976656201910397.post-17367914607083385492013-05-31T09:13:08.877+02:002013-05-31T09:13:08.877+02:00Hi Luca,
Excellent stuff....thanks a lot for shari...Hi Luca,<br />Excellent stuff....thanks a lot for sharing....<br />AlbertoAlbertohttps://www.blogger.com/profile/01805632595255042258noreply@blogger.comtag:blogger.com,1999:blog-7003976656201910397.post-86983970644169311402013-05-13T12:44:29.548+02:002013-05-13T12:44:29.548+02:00Luca, thanks for sharing, this is brilliant work.Luca, thanks for sharing, this is brilliant work.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-7003976656201910397.post-76987219993683811392013-05-12T00:17:17.942+02:002013-05-12T00:17:17.942+02:00It also works on windows client with ansicon :) It also works on windows client with ansicon :) Sayan Malakshinovhttps://www.blogger.com/profile/11087163803358489777noreply@blogger.comtag:blogger.com,1999:blog-7003976656201910397.post-16304338028683689432013-05-11T18:47:18.929+02:002013-05-11T18:47:18.929+02:00Well done guy...!Well done guy...!Database Scenehttps://www.blogger.com/profile/05911995967467360482noreply@blogger.comtag:blogger.com,1999:blog-7003976656201910397.post-86657659262340747492013-05-10T23:30:27.936+02:002013-05-10T23:30:27.936+02:00yes, I think so. I just linked in your post, Frits...yes, I think so. I just linked in your post, Frits' post and Brendan's post on writing "in latency."<br />http://dboptimizer.com/2013/05/10/colored-heat-maps-in-sqlplus/<br />Don't have time to do it myself today, but certainly could be arranged. With enough devices and cgroups defined should be easily possible for readers. :)<br />Kyle Haileyhttps://www.blogger.com/profile/13586511268045480856noreply@blogger.comtag:blogger.com,1999:blog-7003976656201910397.post-83584815580241408882013-05-10T22:27:21.247+02:002013-05-10T22:27:21.247+02:00That is very cool stuff. Another extension would b...That is very cool stuff. Another extension would be to color using how unusual the outliers are relative to the entire GV$EVENT_HISTOGRAM which is taken as representing the latency distribution altogether (after sufficient time under normal workloads) This would be more of a monitoring than diagnostic use case.Anonymoushttps://www.blogger.com/profile/13538337279325457980noreply@blogger.comtag:blogger.com,1999:blog-7003976656201910397.post-64286535178473367552013-05-10T22:21:25.790+02:002013-05-10T22:21:25.790+02:00Hi Kyle, thanks for the comments and for the idea ...Hi Kyle, thanks for the comments and for the idea on how to play around with the output.<br />BTW in his presentation at USENIX/LISA 2010 Brendan Gregg showed that he was able to write short sentences in his NFS latency heat maps, I wonder if that's feasible with Oracle workload too :)Luca Canalihttps://www.blogger.com/profile/06252662329568134677noreply@blogger.comtag:blogger.com,1999:blog-7003976656201910397.post-63157666212751724042013-05-10T22:07:01.088+02:002013-05-10T22:07:01.088+02:00Hi Kevin, thanks for your comments. You are right ...Hi Kevin, thanks for your comments. You are right about the source of the 16-32 ms latency. From the heat maps you can see that on top of that we also have quite a few I/Os with very high latency too. Those systems are getting soon a H/W upgrade BTW (in the mean time they make good example graphs for I/O troubleshooting tools :)<br />I am more than happy to refer to your excellent blog.<br />Luca Canalihttps://www.blogger.com/profile/06252662329568134677noreply@blogger.comtag:blogger.com,1999:blog-7003976656201910397.post-85275269255771842812013-05-10T21:46:18.145+02:002013-05-10T21:46:18.145+02:00Luca,
Very, very good post as usual... althoug...Luca,<br /><br /> Very, very good post as usual... although your 16 to 32ms for mechanical drives is really scary ! Maybe lots of over utilized RAID 5 SATA 7200 RPMs :-) ?<br /><br /> Your point about CPU utilization and how it relates to LFS is spot on. If you don't mind I'd like to refer your readers to a post I have on that very topic:<br /><br />http://kevinclosson.wordpress.com/2007/07/21/manly-men-only-use-solid-state-disk-for-redo-logging-lgwr-io-is-simple-but-not-lgwr-processing/<br /><br /><br /><br />Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-7003976656201910397.post-78835249828563993002013-05-10T20:46:23.310+02:002013-05-10T20:46:23.310+02:00wow - works just like just like that - no install,...wow - works just like just like that - no install, just run the SQL file.<br />Has to be the coolest graphical widget for Oracle performance that I've seen in ages.<br />Works great with https://fritshoogland.wordpress.com/2012/12/15/throttling-io-with-linux/ for playing around with the output.<br />- Kyle Hailey<br />http://dboptimizer.comAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-7003976656201910397.post-56643847737862155922013-05-09T23:47:55.393+02:002013-05-09T23:47:55.393+02:00Crazy!! It works really well, just added to my fav...Crazy!! It works really well, just added to my favorite SQL scripts directory! :-)<br />Thank you for sharing this...<br /><br />@ludovicocaldaraUnknownhttps://www.blogger.com/profile/14760757632604533493noreply@blogger.com