I would like to know, when we can say ASM disks are imbalanced.
Either disks in different size with in ASM diskgroup OR when we have different size ASM diskgroup (with unique disk size in each diskgroup).
Currently one of our Production database (standalone DB) running in ASM storage in that we have multiple diskgroups.
each diskgroup having multiple disks.
- Code: Select all
Group# NAME Size_GB
------ ------------------ -------
1 ARCHLOG 200
3 DATAFILE 250
2 DATAFILE_ORAPRD88 375
6 DATAFILE_ORAPRD86 1000
4 LOG1 20
5 LOG2 20
6 rows selected.
Group# PATH Size_GB
---------- ----------------------------- -------
1 /dev/oracleasm/disks/ARCH003 100
1 /dev/oracleasm/disks/ARCH002 100
2 /dev/oracleasm/disks/DATA004 125
2 /dev/oracleasm/disks/DATA003 125
2 /dev/oracleasm/disks/DATA005 125
3 /dev/oracleasm/disks/DATA002 125
3 /dev/oracleasm/disks/DATA001 125
4 /dev/oracleasm/disks/LOG011 10
4 /dev/oracleasm/disks/LOG003 10
5 /dev/oracleasm/disks/LOG012 10
5 /dev/oracleasm/disks/LOG014 10
6 /dev/oracleasm/disks/DATA009 250
6 /dev/oracleasm/disks/DATA008 250
6 /dev/oracleasm/disks/DATA007 250
6 /dev/oracleasm/disks/DATA006 250
15 rows selected.
As you can see, DATAFILE_ORAPRD88 diskgruop has (3*125G) 3 disks and DATAFILE_ORAPRD86 has 4 disks (4*250),
Three instances are running under DATAFILE_ORAPRD88 diskgruop, and those DB's are very high OLTP are happening.
Last week one of our L4 DBA reported that heavy I/O are happening on this host, especially DATAFILE_ORAPRD88 diskgroup.
finally they suggested like, all disks size in a ASM diskgroup should be of same size, hence he asked us to request more LUNs with 100G each.
and add it to existing diskgroup and asked us to remove those old disks to balance I/O. This is what he is suggested.
As per my knowledge, Oracle recommends ASM diskgroup should have equally sized disks to get I/O balanced!. am I right here?
2. Unix Admin reported that DATAFILE_ORAPRD88 diskgruop having more I/O always, so I took AWR report on all three DBs which is accessing
this diskgroup, and found two DBs are causing huge Physical I/O. upon checking further found same query is running on both the database and its been missing an Index.
what I need to know here is, If I create an Index for that query, I am sure I/O will reduce, if so does I/O on DATAFILE_ORAPRD88 diskgroup will reduce?