Various technologies that are included under MAA are
1.Data Guard
2.RAC
3.Flash Back Recovery
4.ASM
RAC is meant for instance level failures. A multi instance database is nothing but RAC database.
RAC supports :
1.Load balancing
2.Fail over
3.Scalability(adding a node)
Data guard is meant for disk outage/site outage. In the event of any diaster, customer should be in a position to continue his business with the minimal down time. For this oracle has introduced a technology within the database that is called data guard.
Data guard alone or in combination with RAC provides a robost high availability solution.
Even if RAC is in place, in order to have a BCP(Business Continuty Plan) customer has to go for data guard.
There are 2 types of standby databases :
1.Logical standby
2.Physical standby
Logical standby is generally used for reporting purpose and it uses logminer or sqlminer.
If we configure Physical standby with data guard broker then the configuration we call it as data guard.
Data guard broker is an optional component and it is generally used for fast fail over and switch over.
Note : If we want to have BCP, recommanded to configure the physical standby.
Prior to oracle 9i, the configuration mode is called 'delayed protection mode'.
Physical standby is an exact replica of the primary database.
In oracle technology stack, there are 4 different replication technologies.
1.oracle streams
2.oracle data integrator
3.oracle data guard
4.oracle golden gate
Oracle golden gate is a high performance hetrogenous replication tool.
Oracle golden gate reads the data from transaction logs as well as archivelogs and applies the data to a different database.
The site where we configure the physical standby is called 'diaster recovery site'.
In delayed protection mode, archiver takes the responsibility of shipping the generated archive logs to the physical standby.
MRP : Managed Recovery Process
RFS : Remote File Server
In delayed protection mode, there will be lot of data divergance(mismatch) between the primary as well as physical standby.
In delayed protection mode, we cant achieve no data loss solution.
In version 9i, oracle has made lots of architectural changes to the physical standby.
To minimise data mismatch between the primary and the physical standby, oracle has introduced LNS (Logged Network Service).
LNS sucks the data from the redo buffer cache and shifts it to physical standby.
At the physical standby, RFS reads vector by vector and constructs an archive log.
Once the archive log is constructed, MRP applies the data to physical standby.
Note : upto 11g, MRP applies the data to the physical standby, if physical standby is in MRM(Managed
Recovery Mode).
In 11g, MRP applies the data even if the physical standby is in read-only mode. This configuration is called 'active data guard' and requires separate licensing.
Active data guard can be used for real time reporting purpose.
In the above configuration, we cant achieve a no data loss solution.
Note : In order to achieve a no data loss solution, oracle has introduced 'Standby redologs',which needs to be created at the standby side.
The size of standby redologs should be same as the size of primary redologs.
If there is a archive log gap, RFS pings back the archiver(ARCH) to ship the missing archive.
