So best practice?
Jon Gill, September 17, 2008 - 3:16 pm UTC
So are you saying the best practice is to utilize autoextend? and just manage the OS space and go from there? Is this supported as a generally accepted approach/methodology. This is not me asking, but questions I will get asked. I absolutely agree, you can only plan based on historic information. But to beter understand the OS space threshold question, is there a %used for particular types of filesystems that could cause performance degradation and you would want to provide a slight overhead for?
September 17, 2008 - 7:20 pm UTC
I do not believe in best practices actually - if there were a universal "this is best", why would there be other ways?
I like autoextend.
Some people *hate* it.
Is it generally accepted - sure, is it also generally sometimes frowned on - sure, is it a good idea - sure, is it....
This will be a battle you cannot win on this one - there are two diametrically opposed opinions here. I like autoextend with a reasonable maxsize on the datafiles in general (raw would be the exception, although ASM brings it back into being again since ASM makes raw look "not raw"). I prefer my files to grow extent by extent and just keep reserve capacity in the file system. That way you need less overall reserve capacity. If each tablespace had to run at 10% free - and you have tablespaces that grow at different rates (they typically do), you cannot "share" the free space across them. If you let the file system on the other hand run 10% free - any datafile from any tablespace can use that free space.
There is no % "threshold" I am aware of - like I said, it totally depends on your rate of growth.