Veritas InfoScale™ 8.0.2 Storage and Availability Management for Oracle Databases - AIX, Linux, Solaris

Creating an extent balanced file system using SmartTier for Oracle

Any multi-volume file system (MVFS) can become an extent balanced file system, if the storage tier has a chunk size associated with the class. The dbdst_admin command permits the user to define a chunk size for the class.

For example, the following dbdst_admin commands define chunk sizes for the gold and silver storage classes:

$ /opt/VRTS/bin/dbdst_admin -S $ORACLE_SID \
-o definechunk=gold:256K

$ /opt/VRTS/bin/dbdst_admin -S $ORACLE_SID \
-o definechunk=silver:128K

For example:

$ /opt/VRTS/bin/dbdst_admin -S $ORACLE_SID \
-o definechunk=TIER1:256K

The above commands make storage class gold as extent balanced.

Once the chunksize is defined for a storage tier, we can classify any MVFS into this storage tier.

For example, assume that /oradata is the filesystem created on volume-set /dev/vx/dsk/oradg/ora_vset, and contains database datafiles. Let us further assume that datafile names end with extension *.dbf. To define storage class in this MVFS, the following dbdst_classify command is used:

$ /opt/VRTS/bin/dbdst_classify  -S $ORACLE_SID \
-M /dev/vx/dsk/oradg/ora_vset -v  vol1:GOLD,vol2:GOLD,vol3:GOLD

For example:

$ /opt/VRTS/bin/dbdst_classify -S VRTS11r2 \
-M /dev/vx/dsk/dstdatadgot/datavol -v datavol1:FAST,datavol2:MEDIUM

It is important to note that, an MVFS can have multiple storage tiers and that each tier may have a different chunk size. For example, for the same MVFS in the above example, we can define another storage tier using the dbdst_classify command:

$/opt/VRTS/bin/dbdst_classify -S $ORACLE_SID \
-M /dev/vx/dsk/oradg/ora_vset -v vol4:silver,vol5:silver

For example:

$ /opt/VRTS/bin/dbdst_classify -S VRTS11r2 \
-M /dev/vx/dsk/dstarchdgot/archvol -v archvol1:FAST,archvol2:MEDIUM

At this point we have two storage tiers in MVFS /oradata each having different chunksizes. To create the real extent balance, we need to assign a SmartTier policy and to enforce it.

To define and enforce the policy, you could use the following dbdst_preset_policy command:

$/opt/VRTS/bin/dbdst_preset_policy -S $ORACLE_SID \
-d /oradata -P GOLD=*.dbf:SILVER=*.inx

The above example creates a SmartTier policy, assigns the policy to /oradata and enforces the policy. All datafiles of the form *.dbf will be extent balanced in GOLD tier with chunksize 256K and all index files of the form *.inx will be extent balanced in SILVER tier with chunk size 128K.

Another example:

$ /opt/VRTS/bin/dbdst_preset_policy -S VRTS11r2  \
-d /data11r2/VRTS11r2 -P "MEDIUM=temp*.dbf:FAST=*.log"

Assign placement policy file successful on filesystem /data11r2
Begin enforcing filesystem /data11r2 ...
Enforce placement policy successful on filesystem /data11r2

To view the space usage in the /oradata MVFS use the dbdst_show_fs command. For example:

$ /opt/VRTS/bin/dbdst_show_fs -S $ORACLE_SID -m /oradata

For example:

$ /opt/VRTS/bin/dbdst_show_fs -S VRTS11r2 -m /data11r2 -o volume

VOLUME NAME                   SIZE      USED    AVAILABLE       TAG
-----------                   ----      ----    ---------       ---
datavol-b4vset                10240000  3585514 6654486         PRIMARY
datavol1                      1024000   1024000 0               FAST
datavol2                      4608000   2347872 2260128         MEDIUM

When the GOLD or SILVER tier requires more space, we could add extra space by adding new volumes to the respective storage tier using the dbdst_addvol command. For example:

$ /opt/VRTS/bin/dbdst_addvol -S $ORACLE_SID \
-M /dev/vx/dsk/oradg/ora_vset  -v vol7:GOLD

As soon as you add a new volume, the SmartTier policy is enforced and the extents are balanced over the new volume too. This can be viewed by using the dbdst_show_fs command again.

To view detailed extent information about a given file, you can use the fsmap command. For example:

$ /opt/VRTSvxfs/sbin/fsmap -a /oradata/emp.dbf

$ /opt/VRTS/bin/fsmap -a  /oradata/emp.dbf

$ fsmap -a /standby/oradata1_1/slave/system01.dbf |more
 Volume  Extent Type  File Offset  Extent Size   File
  vol1      Data           0      262144      /standby/oradata1_1/slave/system01.dbf
		vol2      Data      262144        8192      /standby/oradata1_1/slave/system01.dbf
  vol3      Data      270336      131072      /standby/oradata1_1/slave/system01.dbf
  vol1      Data      401408      262144      /standby/oradata1_1/slave/system01.dbf
  vol2      Data      663552      180224      /standby/oradata1_1/slave/system01.dbf
		vol3      Data      843776       32768      /standby/oradata1_1/slave/system01.dbf
		vol1      Data      876544       49152      /standby/oradata1_1/slave/system01.dbf
		vol2      Data      925696       65536      /standby/oradata1_1/slave/system01.dbf
		vol3      Data      991232      196608      /standby/oradata1_1/slave/system01.dbf