Database Reference
In-Depth Information
Going back to the initial state of an otherwise idle system, you will find that the number of page tables allocated is
a lot lower and nearly back to the initial 46 MB. Let's review the same with large pages enabled. The maximum value
registered with large pages enabled is shown here:
PageTables: 25228 kb
If you needed more persuasion to allocate large pages for your database, then hopefully this should give you
another good reason to do so.
Deciding about storing files
The storage options available to the database administrator have not changed with Oracle 12.1. As with previous
releases you have the choice of a supported file system or Automatic Storage Management (ASM). With ASM comes
a strong encouragement to use Oracle Managed Files. In this respect it does not matter if the storage is local on the
host, or provided via a SAN or NFS. Most readers will be familiar with the use of ASM and Oracle Managed Files, if you
would like a refresher please refer to Chapter 2 for more information.
If you are not using ASM, you need to think about the location of the seed PDBs file location. Remember from
the section “Common physical structures” that the seed is always present in a Container Database and has a SYSTEM,
SYSAUX, and a TEMP tablespace. Since the seed database cannot be modified after it has been created some planning
is necessary. The seed's data file locations can be configured in the following places:
Using the “seed file_name_convert” clause in the “create database” command.
Setting the “
db_create_file_dest ” initialization parameter to enable Oracle Managed Files.
pdb_file_name_convert ” initialization parameter.
The easiest way to deal with the seed's and all user PDBs' data files is to use Oracle Managed files. Using it you
do not need to worry about any conversion parameters, but this comes at the cost of having slightly less control,
especially when using ASM. You give away some command over file names when using OMF. In addition to that ASM
is not a traditional file system as such where you can use ls , tar , gzip , and other tools you can use on a different
supported file system. Additionally, the data files used by any PDB—seed and user—will include the PDB's globally
unique identifier-GUID-in the file name. The system tablespace has the following fully qualified name in ASM:
Using the “
+DATA/CDB1/DD7D8C1D4C234B38E04325AAE80AF577/DATAFILE/system.259.826623817
If you are not using OMF, then your approach will be different. You will shortly see that the Oracle by default
creates a sub-directory for the seed database:
[oracle@server1 ~]$ find /u01/oradata -type d
/u01/oradata/CDB1
/u02/oradata/CDB1/pdbseed
This directory structure works well for the initial seed database. To make best use of the Pluggable Databases
Feature it can be useful to define the storage location of user-PDBs on SAN attached storage. This way, unplug and
plug operations from one database server to another can be much simpler. All it requires is a remount of that mount
point to another host, followed by plugging the database in. In this scenario the seed database would still go to
/u01/oradata/${ORACLE_SID}/pdbseed , but the user-PDBs could go to /u02/oracle/PDBs/<pdbname>/ for example.
Since the seed is never used for anything other than cloning a PDB, its read/write performance can be ignored most
of the time if you want to place it on slow storage.
 
Search WWH ::




Custom Search