Doc ID:	Note:179312.1
Subject:	A Short Description of HA Options Available in 9i
Type:	BULLETIN
Status:	PUBLISHED

Content Type:	TEXT/PLAIN
Creation Date:	08-MAR-2002
Last Revision Date:	02-JUL-2002

PURPOSE

-------

The purpose of this bulletin is to provide a short description of

the different options available in Oracle 9i to provide high availability.

SCOPE & APPLICATION

-------------------

This bulletin is intended for DBAs who intend to implement a High

Availability solution with Oracle 9i, and need general information about

all available options.

ALTERNATIVES OFFERED BY ORACLE FOR HIGH AVAILABILITY

====================================================

What is high availability?

Systems which provide high availability must be capable of ensuring nearly

full-time availability. Such systems must avoid having single points-of-failure,

by having hardware and(or) software redundancy. In case of failure, tasks

being performed by a failed component are taken over by the backup component.

Oracle offers a wide variety of options providing high availability, on top of

the 9i kernel which already basically ensures a good level of availability.

It is not easy to choose between all of those alternatives, some of which being

usable in combination. This bulletin describes each of the following features,

and discusses the possible combinations of the following different options:

RAC, RAC Guard, Data Guard, Failsafe, and Replication

Considerations about the cost, the needed availability, the downtime allowed in

case of scheduled or unscheduled downtime due to any disaster will help DBAs

to make a choice between the above options.

Oracle Real Application Clusters (RAC)

--------------------------------------

RAC relies on clustered hardware and permits multiple instances to share a

single database:

_________________________________________

| | |

| --------- |

| Node 2 |

| --------- |

| | |

| --------- ------------- --------- |

|-- Node 1 -- Switch/Hub -- Node 3 --|

| --------- Interconnect ---------|

| ------------- |

| | |

| ---------- |

| Node 4 |

| ---------- |

|____________________|___________________ |

| |

___|___ /

/ \______/ Storage Area Network

|\ ______/__

| / \

||\_________/|

|| |

\| database |

| |

\_________/

RAC offers High availability with Transparent Application Failover in which

active sessions on a failed node or instance can be automatically failed over

to surviving instances on other nodes. RAC also provides scalability. The

workload can be spread among all nodes, while maintaining constant response

times. Based on Oracle's Cache Fusion architecture, Oracle9i Real Application

Clusters enables sharing of frequently accessed data across all the servers

in a cluster. If the query request is served by a remote cache, then the block

is transferred across the high speed cluster interconnect from one node's cache

to another. This "fusing of the caches" happens automatically and is

transparent to the application. This transparency is the key technology that

provides the fast, efficient scaling of Oracle9i Real Application Clusters.

This block shipping through the High Speed Interconnect was true in 8i for

read-read pings, it is now also true for write-write pings in 9i. For more

information about RAC and cache fusion, please refer to:

Note 132987.1

High Availability - Real Application Clusters Library Page Index

Real Application Clusters Concepts

Real Application Clusters Installation and Configuration

Real Application Clusters Administration

Real Application Clusters Deployment and Performance

RAC may be combined with:

- Data Guard: Provides protection against data loss

- Advanced Replication: Will enable you to share information with other

hardware platforms or database versions

- Oracle RAC Guard: RAC with additional HA features

Oracle RAC Guard

----------------

Real Application Clusters Guard is an enhanced configuration of Real Application

Clusters. RAC Guard connects all clients to one Oracle instance in normal

operation. In the event of a failure, Oracle's failover and monitoring

software together with a partner's cluster framework will detect the problem

and gracefully switch over clients to the second Oracle instance, ensuring

continued data access. These configurations are tested to ensure there are no

incompatibilities that may disrupt normal operation. RAC Guard has been

designed to quickly and automatically recover from many failures, including

hardware, operating system, or Oracle instance faults. It provides fast and

bounded recovery, enforced primary/secondary node access, automatic reconnect

of failed sessions, and capture of diagnostics data after a failure.

Example of architecture:

------------ ------------ ------------

Node 1 Node 2 Node 3

_______ ________ _______

Primary Primary Primary

__A____ __ B____ __C____

| | |

------------ ------|------ -------|----

| | |

| V |

| ------------------------------------------ |

| | Node 4 ----------- | |

| | Secondary | |

| | B | |

| | ----------- | |

| | | |

| | ----------- ----------- | |

|--|--> Secondary Secondary <----|-|

| A C |

| ----------- ----------- |

| |

------------------------------------------

RAC Guard may be combined with:

- Data Guard: Provides protection against data loss

- Advanced Replication: Will enable you to share information with other

hardware platforms or database versions

For more information about RAC Guard, please refer to:

Oracle Real Application Clusters Guard Administration and Reference Guide

For more information about RAC Guard on Windows, please refer to:

Oracle Real Application Clusters Guard Installation Guide

Oracle Real Application Clusters Guard Concepts and Administration Guide

Oracle Services for MSCS Error Messages

Oracle Failsafe

---------------

Oracle Failsafe is a core feature of Oracle9i, Oracle iAS, and Oracle

Applications Release 11i that provides high availability for e-business

solutions deployed on Windows clusters. Oracle Fail Safe release 3.2 works

with Microsoft Cluster Server to ensure that if a failure occurs on one cluster

node, then the Oracle databases and applications running on that node will

automatically fail over to a surviving node, therefore minimizing downtime.

Oracle Failsafe is optimized for Windows customers with database and application

workloads that can be handled by a single system. Oracle Failsafe solutions can

be deployed on Windows NT and Windows 2000 clusters. It supports up to 4

servers in a cluster with Windows 2000.

Supported products include:

Oracle Databases (Standard and Enterprise Editions of Oracle9i and Oracle8i)

Oracle Applications release 11i

Oracle iAS components, including:

Oracle Forms Services

Oracle Reports Services

Oracle HTTP Server

Oracle Intelligent Agent

Oracle Service for Microsoft Transaction Server (for Oracle8i release 8.1.7)

Oracle Failsafe may be combined with:

- Data Guard: Provides protection against data loss

- Advanced Replication: Will enable you to share information with other

hardware platforms or database versions

To know more about Failsafe please refer to:

Note 146388.1

High Availability - Oracle Fail Safe Library Index

Oracle Fail Safe Installation Guide

Oracle Fail Safe Concepts and Administration Guide

Oracle Services for MSCS Error Messages

Oracle Data Guard

-----------------

Oracle9i Data Guard provides physical and logical standby database

(available in 9.2+ only) protection and disaster recovery features to help you

to survive mistakes, corruptions, and other disasters that may destroy a

database. Data Guard broker protects critical data by automating the creation,

management, and monitoring aspects of a standby database environment.

In 8i, it is possible to configure a physical standby database, which is a

mirror of a primary database where automatically shipped archived redo logs are

continuously being applied. If the primary database goes down, the standby

database will take over the primary role and the interruption of service will

be minimized.

In 9.2, the logical standby database will be introduced where archived redo logs

are mined and the extracted SQL statements are applied to the logical standby.

An advantage of this feature is the capability of reading and writing

(especially to create indexes or do some maintainance) to the logical standby

while applying logs through SQL statements.

Example of architecture:

Primary Site Standby Site

------------ ------------

-------------- ________ ________ --------------

| | / \ / \ | |

| | |\________/| |\________/| | |

| | | | | | | |

| Oracle | \________/ \________/ | Oracle |

| Data | | Data |

| Guard |Online Archived Archived | Guard |

| Broker |Redo Logs Redo Logs Redo Logs | Broker |

| | _____ _____ _____ | |

| |/ \ / \ / \ | |

| |\_____/| |\_____/| |\_____/| | |

-------------- | | | | | --------------

| | |\_____/| |\_____/|

| | | | | |

\_____/ \_____/ \_____/

|__________^ ^

| Local |

| Archiving |

|_______________ /\_________|

\/ Remote

Archiving

Via

Oracle Net

Oracle Data Guard may be combined with:

- Advanced Replication: Will enable you to share information with other

hardware platforms or database versions

- RAC and RAC Guard: Will provide additional HA and scalability features

For more information about Data Guard, please refer to:

Note 146387.1

Backup and Recovery - Standby Databases and Data Guard Library Index

Oracle9i Data Guard Concepts and Administration

Oracle9i Data Guard Broker

Oracle Replication

-------------------

Replication is the process of copying and maintaining database objects in

multiple databases that make up a distributed database system. There are three

ways of replicating data :

- Read-only materialized views: One master table and multiple remote copies

being refreshed at intervals by pulling changes from the "master site" to

the "slave site".

- Updatable materialized views: One master table and multiple partial or entire

copies on snapshot sites (updatable).

- N-way masters replication: Multiple master tables being updatable from all

sites. Changes applied at one site are captured and stored locally before

being forwarded and applied at each of the remote locations.

In addition to providing users with fast, local access to shared data,

applications enjoy increased availability because alternate data access

options exist. Even if one site becomes unavailable, users can continue to

query or even update the remaining locations.

Oracle Replication is suitable for applications which typically require data to

be periodically synchronized between central systems and very large numbers of

small, remote sites often operating in a disconnected manner. Oracle provides

a sophisticated conflict detection mechanism and a comprehensive set of

automated conflict resolution routines to ensure data convergence throughout

the replicated environment.

Oracle Replication facilitates information sharing across multiple hardware

platforms and database versions.

Replication works at an object level, whereas other options work at a database

level. Please note that some operations like adding objects in a replication

group, require quiescing the replication group, making the replicated objects

unavailable. A quiesced state prevents users from executing any transactions

against a replicated object in the quiesced master group. Oracle Replication

has been enhanced in 9i and now it is no longer necessary to quiesce a

replication group when adding or removing a master database.

Example of architecture:

Multiple master sites

for high availability

______ Synchronisation

/ \ of data over

|\______/| Internet ___

| | <----------> / \

---->| Master |<--- |\___/|

| | | | | |

| \______/ | \___/

| |

| | ___

_V____ ___V__ / \

/ \ / \ <----------> |\___/|

|\______/| |\______/| | | Remote sites

| | | | \___/ operating in

| Master | | Master | disconnected

| | | | manner

\______/ \______/ ___

\ / <----------> / \

-<----------->- |\___/|

| |

\___/

Oracle Replication may be combined with:

- Data Guard: Provides protection against data loss

- RAC and RAC Guard: Will provide additional HA and scalability features

For more information about replication, please refer to:

Note 132256.1

Distributed - Advanced Replication Library Index

Oracle9i Replication

Oracle9i Replication Management API Reference Release 1 (9.0.1)

Conclusion:

----------

When making a decision about high availability, it will be important to weigh

the costs vs. the cost of downtime, the various features available, and the

implementation of these features.