Hope can get some input regarding the setup above.
Understand that Oracle best practise is to have all LUNS in a diskgroup to have same sizing and performance characteristic.
Currently one of the system is running on file system with 70+ mountpoints, those mountpoins are built from LUNS from different storage pools (Mix of EMC/HP).
Now there is a plan to migrate the DB into ASM by stage by reusing back the current LUNS.
What impact would I expect if I were to combine all current mountpoints(LUNS) and assign it into one single diskgroup?
As Oracle doc and support always ask us to use the LUNS with same size and performance characteristic, but that wouldn't be ideal case as when new storage coming in, it's different from whatever it's currently using.
So what puzzling me is it's running ok on file system with mixture of different storage, but if we were to mix them together in a diskgroup, will the performance remain the same as of current running file system setup?(reading/writing still travel across all the storage pools)
Here is the example:
/u01/,/u02/,/u03/,/u04/ - 4 LUNS from EMC
/u05,/u06/,/u07,/u08/,/u09,/u10 - 6 LUNS from HP
all having same size.
The plan is to temporary move out the datafile from file system to temp storage and destroy the existing mountpoint, then reuse/assign these 10 LUNS into a single diskgroup.
Would like to know the feasibility of doing this. Thanks
First if you can create new LUNS then you can just migrate from non-ASM to ASM.
Have the same type of LUNS is the best practice I have seen performance issue where mixed IOPS speeds for different LUNS causing problems. The LUNS size is a must they SHOULD be the same size always if you where going to move lets say from 50G LUNS to 100G then you could temporarily add the 100G LUNS then remove to 50G LUNS but you would never mix the different size of LUNS in one diskgroup.
let me know if this helps.