Bug 9829397  Excessive CPU and many "asynch descriptor resize" waits for SQL using Async IO

Affects:

Product (Component)	Oracle Server (Rdbms)
Range of versions believed to be affected	Versions >= 11.2 but BELOW 12.1
Versions confirmed as being affected	11.2.0.2 11.2.0.1
Platforms affected	Generic (all / most platforms affected)

 It is believed to be a regression in default behaviour thus:
   Regression introduced in 11.2.0.2

Fixed:

This issue is fixed in

12.1.0.1 (Base Release) 11.2.0.3 (Server Patch Set) 11.2.0.2.5 Database Patch Set Update 11.2.0.2.5 Grid Infrastructure Patch Set Update (GI PSU) 11.2.0.2 Bundle Patch 13 for Exadata Database 11.2.0.2 Patch 13 on Windows Platforms 11.2.0.1 Patch 15 on Windows Platforms

Description

Some queries in 11.2 may exhibit higher CPU usage than earlier
releases with many "asynch descriptor resize" waits occurring
compared to the same SQL in earlier releases. 
 
Rediscovery Notes:
 Async IO is in use.
 The total time waiting for "asynch descriptor resize" is
 typically very small but with very high counts. The high
 wait count indicates many resizes of the number of AIO 
 descriptors unnecessarily wasting CPU. 

Workaround Disable async IO. 
 eg: Set DISK_ASYNCH_IO = false    <<<< 禁用异步IO
 
References:
 For more information about "asynch descriptor resize" see the following: Note:1273748.1 High Numbers of 'asynch descriptor resize' waits Note:1081977.1 Details of the "asynch descriptor resize" wait event.
 

Getting a Fix Use one of the "Fixed" versions listed above (for Patch Sets / bundles use the latest version available as
  contents are cumulative - the "Fixed" version listed above is
  the first version where the fix is included) or Click here for suggestions on how to get a fix for this issue 

References

Bug:9829397 (This link will only work for PUBLISHED bugs)
Note:245840.1 Information on the sections in this article