Re: troubleshooting slow I/O performance.

From: Stefan Koehler <>
Date: Wed, 9 May 2018 21:26:51 +0200 (CEST)
Message-ID: <>

Hey Chris,
there is a pretty old paper called "Useful Constants for the Oracle Performance Analyst" by Cary Millsap and Jeff Holt which mentions 10 ms per single-block disk read (PIO) as max. tolerated latency per event.

However you usually see much lower latency per event (basically what you have expected) as storage sub-systems (even the cheaper mid-range ones) have read/write caches to support the spinning disks. So I would not consider 10 ms as "normal" as long as I am not sure about the root cause of it.

Best Regards
Stefan Koehler

Independent Oracle performance consultant and researcher Website:
Twitter: _at_OracleSK

> Chris Stephens <> hat am 9. Mai 2018 um 15:48 geschrieben:
> thanks for everyone's input on this. the 10ms I/O times are relatively constant on this disk group.  we were just looking to make sure those disks are performing as expected.  I have always considered 5ms single block I/O response times for spinning disks as typical and when I saw 10ms I thought maybe something was up. not sure how i got that 5ms number stuck in my head.
> Luckily we also have a disk group on SSD's that performs far better.
> thanks again everyone.  even you and your criticisms mladen. :)

Received on Wed May 09 2018 - 21:26:51 CEST

Original text of this message