Ever since Oracle released their database, DBAs have been finding it increasingly difficult to manage large numbers of database files while keeping them performing optimally. In addition, maintaining the redundancy of those data files at their expected levels poses a challenge. Over the decades, many hardware vendors have provided various storage solutions, yet these solutions have yet to meet the performance levels required by data files today. To meet these needs, Oracle developed the exclusive software solution, ASM. ASM is an integrated file system and volume manager expressly built for Oracle database files. So, what is the big advantage ASM provides? Simply put, DBAs can manage a small of number of disk groups rather than a large number of data files.

A disk group is a set of disks managed together as a unit. How can this optimize performance? ASM automatically creates data files and uses Stripe And Mirror Everything (SAME) technology to stripe and mirror the data files evenly across the disks. Moreover, ASM stripes data at a file level using an allocation unit of 1MB, unlike at volume level. ASM also mirrors the data allocation units in different disks.

In the event of a disk failure, ASM is able to recover the data automatically by evenly redistributing the data over the remaining disks in the disk group. ASM can also fully recover the redundancy of data depending upon the remaining space in the disk group.

Nowadays, disks are so advanced that it is highly unlikely all disks in a disk group will fail at the same time. Since ASM can tolerate a couple of disk failures, the disk group will not lose its data. More surprisingly, most of the failure situations are taken care of by ASM automatically. The DBA can even add or remove a disk from a disk group while the database is online. ASM works well with single or clustered databases too. Multiple databases can also use the same ASM storage arrays at the same time.

ASM internally uses Oracle’s standard file architecture, called Oracle Managed Files (OMF), to create and delete files. It eliminates the need for the DBA to directly manage the operating system files within the Oracle database. ASM inherently supports very large files because any disk group can have a maximum of 10,000 disks. ASM is not only making life easier for the DBA, but also cuts down on overall resource costs and maximizes disk performance.

Configuration below uses IBM System Storage DS8000

Set up storage pool (extent pool)

To create the extent pool:
1. Create an array.
dscli> mkarray -dev IBM.1750-13AB73A -raidtype 5 -arsite IBM.1750-13AB73A/S1

2. Create one fixed block from one array.
dscli> mkrank -dev IBM.1750-13AB73A -array A0 -stgtype fb

3. Create a fixed block storage type extent pool.
dscli> mkextpool -rankgrp 0 -stgtype fb ora_RAC

4. Assign an unassigned rank to a extent pool.
dscli> chrank -extpool p0 r0

5. Display a list of array sites and status information.
dscli> lsarraysite
arsite   DA Pair     dkcap (10^9B) State     Array
====================================================
IBM.1750-13AB73A/S1 IBM.1750-13AB73A/0 146.0 Assigned
IBM.175013AB73A/A0

6. Display a list of defined ranks in a storage image and status information.
dscli> lsrank
ID   Group   State   datastate   Array RAIDtype   extpoolID   stgtype
=====================================================
IBM.1750-13AB73A/R0 0 Normal Normal IBM.1750-13AB73A/A0 5
IBM.1750-13AB73A/P0 fb

List the extent pool
dscli> lsextpool
Name   ID   stgtype   rankgrp   status   availstor (2^30B)
%allocated  available  reserved  numvols
====================================================
ora_RAC   IBM.1750-13AB73A/P0 fb   0   below   48   81
48  0  28

Make volume group and LUNs

To create volume group and LUNs:
1. Create a volume group in a storage image.
dscli> mkvolgrp -type scsimask Aix_oracle
creates volumegroup named Aix_oracle and assigns an
identifier in this case the identifier is v12
2. Create an open systems fixed block volume in a storage image.
dscli> mkfbvol -extpool p0 -cap 1 –volgrp v12 -name
oracle_#h 0210-0213
1GB luns for CRS and vote disks (4 total)
dscli> mkfbvol -extpool p0 -cap 10 –volgrp v12 -name
oracle_#h 0410-0419
10g luns for ASM disks (10 total)
3. List the resulting volume group and its members.
dscli> showvolgrp v12
Name Aix_oracle
ID IBM.1750-13AB73A/V12
Type SCSI Mask
Vols 0210 0211 0212 0213 0410 0411 0412
0413 0414 0415 0416 0417 0418 0419

Use SDDPCM to manage the fiber connections

SDDPCM provides commands to display the status of adapters used to access managed devices, to display the status of the devices that the device driver manages, or to map supported storage MPIO devices or paths to a supported storage device location.

1. Display the SDDPCM path information.
$ pcmpath query device
DEV#: 2 DEVICE NAME: hdisk2 TYPE: 2107900 ALGORITHM: Load
Balance
SERIAL: 75023012100
==================================================
Path# Adapter/Path Name State Mode Select Errors
0 fscsi0/path0 OPEN NORMAL 171652 0
1 fscsi0/path1 OPEN NORMAL 171502 0
2 fscsi1/path2 OPEN NORMAL 171440 0
3 fscsi1/path3 OPEN NORMAL 171543 0

Create Cluster disks using DS8000 Volumes

1. Create the shared volume using DS8000
2. Assign these volumes for the CRS OCR and vote disks.
3. Create the major and minor numbers and the character device of those disks.
$ chdev -l hdisk2 -a reserve_policy=no_reserve
$ chdev -l hdisk3 -a reserve_policy=no_reserve
$ mknod /dev/ocr_disk c 25 7
$ mknod /dev/vote_disk c 25 9
4. Set proper permission and access level on those shared volumes.
$ chown oracle:dba ocr_disk
$ chown oracle:dba vote_disk
$ chmod 660 ocr_disk
$ chmod 660 vote_dis
5. During the CRS installation, select the external redundant disks to be used as the OCR and voting disks.

Prepare DS8000 Volumes for ASM

1. Create the shared volume using DS8000
2. Assign these volumes for the ASM disks.
3. Create a major and minor numbers and a character device of those disks.
$ chdev -l hdisk6 -a reserve_policy=no_reserve
$ chdev -l hdisk7 -a reserve_policy=no_reserve
$ mknod /dev/asm_disk1 c 25 2
$ mknod /dev/asm_disk2 c 25 13
4. Set proper permission and access level on those ASM disks.
$ chown oracle:dba /dev/asm_disk1
$ chown oracle:dba /dev/asm_disk2
$ chmod 660 /dev/asm_disk1
$ chmod 660 /dev/asm_disk2