Re: Expert Cache

Of course, this has the best effect when file access is basically 
consistent, because it affects all IO requests, not table by table. 

Expert cache works object by object. It tries to recognize the pattern of 
access to each object
and optimize paging activity for that object.

Leaving it *FIXED uses a 4KB request size no matter what is going on.

Each object has a default transfer size, which is not necessarily 4KB.
This is what it falls back to when Expert cache is off.
For example, QUSCRTUS API (Create user space) allows to set the transfer 
size.
Look for optional parameter group 3.
        
http://publib.boulder.ibm.com/infocenter/iseries/v5r4/topic/apis/quscrtus.htm


    Alexei Pytel
always speaking for myself


"We have all reasons to assume that our assumptions are not unreasonable."