user4721711 wrote:post SQL that populates the table.
Currently building a new 188.8.131.52 RAC on OL6.
I am using ASM on top of a SAN that is exposing 4x2TB LUNS to the 3 nodes, as the DATA disk group.
The nodes are communicating over 10gb isci to the SAN.
When creating a test table space I am only seeing about 3-5 mb/s write throughput from Oracle. When testing with ioZone tool to write 8k block size files on to the drives
I am seeing ~115 mb/s throughput.
That should show the results for a "typical" database workload. Ther results are often surprising. You may find that a daabase on your laptop out performs something an a massive SAN.
var al number var mi number var mm number exec dbms_resource_manager.calibrate_io(num_physical_disks=>4,max_iops=>:mi,max_mbps=>:mm,actual_latency=>:al) print al print mi print mm
Output from one of the runs showing 24 mb/s which is 6mb/s for each of the 4 disks in the group.ASM is not RAID and functions on the concept of disk failure groups and available free space.
Also try change filesystemio_options=setall to filesystemio_options=asynch :
This statement has now been running for about 20hrs CREATE BIGFILE TABLESPACE TEST_TBS DATAFILE '+DATA' SIZE 2T and I/O throughput remains steady at ~10-12mb/s