Oracle9i Real Application Clusters Concepts Release 1 (9.0.1) Part Number A89867-02 |
|
This section describes the new features, variables, parameters, views, and installation procedures introduced by Real Application Clusters.
Real Application Clusters is a new architecture offering scalability and high availability features that exceed the capabilities of previous Oracle cluster-enabled software releases. Real Application Clusters introduces a new phase of Cache Fusion, a breakthrough technology that guarantees cache coherency among multiple cluster nodes without incurring disk I/O costs. The first phase of Cache Fusion was introduced in Oracle8i to improve read/write concurrent data access. Real Application Clusters extends that capability to optimize read/read, read/write, and write/write concurrency among multiple cluster nodes. The new features introduced in Real Application Clusters greatly enhance Oracle cluster software performance and scalability.
Real Application Clusters introduces a number of significant improvements. These improvements are divided into several categories:
Real Application Clusters introduces a number of terms to more accurately reflect new functionality:
db_name.conf
file on UNIX and in the Registry on Windows NT platforms.
SYSTEM
, USERS
, TEMP
, and other tablespaces.
Oracle9i Real Application Clusters Installation and Configuration for details on changes to configuration types, raw partitions, templates, Oracle Universal Installer, and DBCA options.
See Also:
Oracle9i
Real Application Clusters Installation and Configuration and the Release 1 (9.0.1) README file for details on raw partition tablespace size requirements
See Also:
See Also:
Oracle9i Database Installation Guide for Windows for information on OSD software installation
See Also:
See Also:
Oracle9i Real Application Clusters Installation and Configuration for information on the DBCA |
Oracle Corporation recommends that you use server parameter files when you implement Real Application Clusters.
Oracle Corporation also recommends that you modify the default file location for server parameter files for Real Application Clusters environments if you are using raw devices.
SYS
, SYSTEM
, and SCOTT
, expire upon installation. To use these names, you must explicitly unlock them.
See Also:
Chapter 2, "Real Application Clusters Architecture" for a description of Dynamic resource remastering
See Also:
Real Application Clusters employs resource affinity. Resource affinity is the use of dynamic resource remastering to move the location of the resource masters for a database file to the instance where operations are most frequently occurring.
See Also:
Chapter 2, "Real Application Clusters Architecture" for a description of resource affinity |
_startup
(Transport Network Services (TNS) entries that had been required for each instance are now obsolete.
ALTER_SYSTEM
, you can dynamically redirect an instance's undo processing from one undo tablespace to another.
See Also:
Oracle9i Real Application Clusters Administration for information on the Automatic Undo Management mode |
See Also:
Oracle9i Real Application Clusters Administration for information on Oracle Enterprise Manager |
See Also:
Chapter 8, "Real Application Clusters Storage Considerations" and Oracle9i Real Application Clusters Installation and Configuration for information on server parameter files |
See Also:
Oracle9i Real Application Clusters Deployment and Performance for details on the statistics charts |
See Also:
Chapter 2, "Real Application Clusters Architecture" for a description of the diagnosability daemon |
See Also:
Oracle9i Real Application Clusters Administration for information on the Oracle Intelligent Agent and the DBCA. |
Beginning with Release 1 (9.0.1), Oracle Parallel Fail Safe (OPFS) was replaced by Oracle Real Application Clusters Guard. It also became a feature of Real Application Clusters on UNIX platforms. Oracle Real Application Clusters Guard has these advantages:
See Also:
Chapter 10, "High Availability Concepts and Best Practices" for information on high availability and failover recovery |
To improve failover performance in Primary/Secondary instance configurations, use the DBMS_LIBCACHE
package to transfer information from the library cache of the primary instance to the library cache of the secondary instance.
See Also:
|
SHUTDOWN TRANSACTIONAL LOCAL
command was introduced as an option to the existing SHUTDOWN TRANSACTIONAL
command. This new command can be used to prevent new transactions from starting locally, and to perform an immediate shutdown after all local transactions have completed. With the new command, you can gracefully move all sessions from one instance to another by shutting down selected instances transactionally.
See Also:
Chapter 10, "High Availability Concepts and Best Practices" and Oracle9i Real Application Clusters Administration for information on the Quiesce Database feature |
A number of SQL scripts were replaced in Real Application Clusters.
utlopslt.sql
script is replaced by utlclust.sql
catparr.sql
script is replaced by catclust.sql
ops.sql
script is replaced by clustdb.sql
The TRACE_ENABLED
parameter was added in Real Application Clusters. When enabled, this dynamic parameter provides low overhead memory tracing. It is enabled by default.
See Also:
Oracle9i Real Application Clusters Deployment and Performance for information on the |
There are three new cluster software parameters in Real Application Clusters.
OPS_INTERCONNECTS
parameter is replaced by CLUSTER_INTERCONNECTS
. Note that with Sun Clusters configurations the interconnect High Availability feature is not available.
PARALLEL_SERVER
parameter is replaced by CLUSTER_DATABASE
.
PARALLEL_SERVER_INSTANCES
parameter is replaced by CLUSTER_DATABASE_INSTANCES
.
Oracle9i Real Application Clusters Deployment and Performance for additional information on how to use the new
See Also:
CLUSTER_INTERCONNECTS,CLUSTER_DATABASE, and CLUSTER_DATABASE_INSTANCES
parameters.
The following parameters are obsolete in Real Application Clusters:
The following new views have been added to Oracle cluster software for Real Application Clusters. These views track the current and previous master instances and the number of re-masterings of enqueue (V$HVMASTER_INFO
), global cache (V$GCSHVMASTER_INFO
), and global cache resources belonging to a file accessed frequently by a single instance (V$GCSPFMASTER_INFO
):
V$HVMASTER_INFO
for Global Enqueue Service resources
V$GCSHVMASTER_INFO
for Global Cache Service resources except those belonging to files mapped to a particular master.
V$GCSPFMASTER_INFO
for Global Cache Service resources belonging to files mapped to a particular master.
The following views have been replaced in Real Application Clusters:
V/GV$DLM_MISC
is replaced by V/GV$GES_STATISTICS
V/GV$DLM_LATCH
is replaced by V/GV$GES_LATCH
V/GV$DLM_CONVERT_LOCAL
is replaced by V/GV$GES_CONVERT_LOCAL
V/GV$DLM_CONVERT_REMOTE
is replaced by V/GV$GES_CONVERT_REMOTE
V/GV$DLM_ALL_LOCKS
is replaced by V/GV$GES_ENQUEUE
V/GV$DLM_LOCKS
is replaced by V/GV$GES_BLOCKING_ENQUEUE
V/GV$DLM_RESS
is replaced by V/GV$GES_RESOURCE
V/GV$DLM_TRAFFIC_CONTROLLER
is replaced by V/GV$GES_TRAFFIC_CONTROLLER
V/GV$LOCK_ELEMENT
is replaced by V/GV$GC_ELEMENT
V/GV$BSP
is replaced by V/GV$CR_BLOCK_SERVER
V/GV$LOCKS_WITH_COLLISIONS
is replaced by V/GV$GC_ELEMENTS_WITH_COLLISIONS
V/GV$FILE_PING
is replaced by V/GV$FILE_CACHE_TRANSFER
V/GV$TEMP_PING
is replaced by V/GV$TEMP_CACHE_TRANSFER
V/GV$CLASS_PING
is now replaced by V/GV$CLASS_CACHE_TRANSFER
V/GV$PING
is replaced by V/GV$CACHE_TRANSFER
The following sections describe the changes to previous Oracle cluster software products.
Beginning with release 8.1.7, Oracle Parallel Server had the following changes:
Some of the raw partition tablespace size requirements changed for Oracle Parallel Server release 8.1.7 as shown in the following table. These tablespaces require slightly greater capacities than the values that were published in release 8.1.6 documentation.
Create a Raw Device for | With File Size |
---|---|
SYSTEM |
|
TEMP |
|
DRSYS |
The process of configuring Oracle Parallel Server to connect to secondary instances was simplified for release 8.1.7. Use the INSTANCE_ROLE
parameter in the Connect Data portion of the connect descriptor to configure explicit secondary instance connections.
The procedures for connecting Recovery Manager (RMAN) to a target database in an Oracle Parallel Server cluster changed for release 8.1.7. For more information refer to Oracle9i Recovery Manager Reference.
OPS_INTERCONNECTS
provides information about additional cluster interconnects for use in Oracle Parallel Server environments. Oracle uses the information from this parameter to distribute traffic among the various interfaces. You would normally use OPS_INTERCONNECTS
when a single interconnect is insufficient to meet the bandwidth requirements of large Oracle Parallel Server databases.
OPS_INTERCONNECTS
is an optional parameter. If you do not set it, then the current semantics that determine the appropriate interconnect for Oracle Parallel Server internode communication are preserved.
The syntax of the parameter is:
OPS_INTERCONNECTS =
if1:
if2:...:
ifn
Where ifn is an IP address in standard dotted-decimal format, for example, 144.25.16.214
. Subsequent platform implementations may specify interconnects with different syntaxes.
Note that when you set OPS_INTERCONNECTS
in Sun Cluster configurations, the interconnect High Availability features are not available. In other words, an interconnect failure that is normally unnoticeable would instead cause an Oracle cluster failure.
Beginning with release 8.1.6, Oracle Parallel Server had the following changes:
With the Primary/Secondary Configuration feature you can implement a basic high availability configuration using the Primary/Secondary Instance feature. This feature serves two-node Oracle Parallel Server environments. The primary instance on one node accepts user connections while the secondary instance on the other node only accepts connections when the primary node fails.
The following statistics were added:
The following statistics became obsolete:
The default setting for GC_ROLLBACK_LOCKS
is "0-128=32!8REACH
" This protects rollback segments 0 through 129 with locks.
Oracle automatically sets values for LM_LOCKS
and LM_RESS
based on settings in your initialization parameter files.
The LM_PROCS
parameter became obsolete in release 8.1.6.
|
Copyright © 1996-2001, Oracle Corporation. All Rights Reserved. |
|