Ryan We did this excercise few months back. Basically, we tried a combination of different LUN sizes and RAID levels to ascertain the combination that worked well (in terms of I/O throughput) for our SAN environment. The factors we considered for LUN sizes were manageability/recoverability/growth rates/snapshot storage space required. Our implementation is 10g RAC ASM/SAN on 64-bit linux platform. Rajeev On 9/15/07, ryan_gaffuri@xxxxxxxxxxx <ryan_gaffuri@xxxxxxxxxxx> wrote: > > I was talking to an Oracle guy who also works with SANs. This is what he > told me. For optimal performance do the following > > 2 redo log groups each with dedicated spindles > 2 archive logs each with dedicated spindles > > data data files have their own dedicates spindles with several lines and > spread the data files out > index data files do the same thing but have their own dedicated spindles > > control files can be placed on the same spindles as data files, however, > check point activity on systems with large numbers of data files especially > when doing hot backups may require control files having their own dedicated > spindles. > > What is your opinion? Is there any way to estimate a good starting point > to test LUN sizes for performance? Not sure what sizes I should ask for to > begin my performance tests. I was thinking of running bench marks with > several different sized LUNs. > -- > //www.freelists.org/webpage/oracle-l