This is an excerpt from the bestselling book
Oracle Grid & Real Application Clusters. To get immediate
access to the code depot of working RAC scripts, buy it
directly from the publisher and save more than 30%.
On UNIX/Linux-based systems,
where relink of the Oracle executables is possible, having a single
instance running from a RAC-disabled Oracle home installation is
possible, if improbable, if a one-node cluster with RAC installation
was performed but later the RAC feature was disabled by unlinking it
from the Oracle binary before creating the single instance database.
An alternative method is to select the local, non-cluster option on
the Node Selection screen in OUI to create a non-RAC-enabled single
instance home on a cluster. The following technique can be used to
convert this special type of single instance database into a RAC
1. On the cluster node where the
single instance database is running, create a duplicate template of
the database as described in previous sections using DBCA.
2. Change the working directory
to the lib subdirectory in the rdbms directory under the Oracle
3. Relink the Oracle binary
through use of the following commands:
make -f ins_rdbms.mk ioracle
At this point, the normal
conversion procedure for a RAC enabled Oracle home can be used.
Conversion of a Single
Instance on a Cluster Running from Non-Cluster Installed Oracle Home
This situation is only possible
if the Local Installation option was specified on the OUI Specify
Hardware Cluster Installation screen during an Oracle Database 10g
To covert this type database to
a RAC database, perform the procedures described under these
* Duplicate the Original Single
* Perform the Pre-Installation
* Setting up the Cluster
* Install Oracle Database 10g
Software with RAC. In this step, make sure that a new Oracle home
other than the one from which the single instance database was
running is selected.
After completing the conversion,
the following items should be reviewed as covered in other chapters
of this book:
* Use of load balancing and TAF
* Considering the use of locally
managed tablespaces instead of dictionary managed tablespaces to
reduce contention and manage sequences
* Properly configuring an
interconnect for using automatic segment space management and for
using SRVCTL to administer multiple instances
* Sizing of database buffer
cache and shared pool: Database buffer cache and shared pool
capacity requirements in RAC are greater than those in single
instance Oracle databases. Therefore, increase the size of the
buffer cache by at least 10% and the size of the shared pool by at
* If ASM is used for the file
systems, use RMAN for backup.
Moving HA Cluster to a RAC
Many enterprises currently have
the High Availability (HA) clusters in their data centers. For
example, a system may have a two node VCS based failover cluster. It
is an Oracle single instance that is a part of one of the VCS
Service groups. This section will briefly examine the issues and
procedures involved in moving a HA based Oracle instance to
multi-node RAC parallel database.
Procedure to Move the VCS to
Veritas DBE/AC RAC
For this section, assume that
there is a HA cluster, or so called Failover Cluster, supporting
Oracle 9i Standalone Database. The single instance is implemented
with VCS cluster framework. This is currently in an Active/Passive
To take advantage of the
multi-node, scalable architecture of the parallel database, which is
a 10g RAC, it must be migrated to the Oracle Database 10g RAC using
the Veritas DBE/Advance Cluster software.
The following procedure can be
used to undertake such migration.
A look at the current
environment shows that there are two servers currently supporting a
two-node HA cluster. All the Oracle Database Files are already on
shared storage volumes. The database files are file system based
files. Optionally, they may have been using the Quick I/O facility.