MAA (MAXIMUM AVAILABLE ARCHITECTURE)
To overcome instance level failovers oracle as introduced multi instance database , which is nothing but RAC database .
RAC supports two types of failover’s:
1 . session failover
- select failover
- There is no concept of DML failover in RAC system.
- Various technologies that are including under MAA are :
- DATA GUARD
- FLASHBACK TECHNOLOGY
Oracle as redefined the term High Availability to MAA
Data guard is meant for DISK OUTAGE (or) SITE OUTAGE. Data guard is a disaster Recovery solution. Data guard alone are in combination with RAC provides a Robost high availability solution.
- Prior to 9i, data guard is called are standby
- In version 9i, oracle as made lots of architectural changes to its high availability technology and rename the technology as dataguard.
There are two types of stand by
- Logical standby
- Physical standby
- Logical standby uses the concept of SQL apply and it is generally used for Reporting purpose.
- Data guard are physical stand by there is also called as Replicating Technology. It is Replicating the data.
- If we want have a business continuity plan and disasters continuity solution, we need to configure physical stand by.
- If we configure physical standby with “ data guard broker” then we call the configure as data guard .
- Prior to 9i configuration mode is called delayed production mode.
- In delayed production mode , archiver background process takes the responsibility of shipping the generated archivelogs to the physical standby.
- RFS (remote file server ) reads vector by vector and constructs Archive logs .
- Then constructed archivelogs will be picked by MRP (manage Recovery process) and apply’s the data to the physical stand by if it runs in MRM mode(Manage Recovery mode)
- Physical standby may run either in MRM mode or READ ONLY MODE.
- In the industry , Some times we open the physical standby in READ ONLY MODE to generate the resource intensive reports.( off loading the primary).
- In the above architecture there will be lot of “ data divergence” (miss match) between the primary and physical standy
- In the industry we are going to maintain the physical stand by exatly replic off primary database. We are going to the physical stand by in remote geographical.
- In order to minimize data divergence , oracle as introduced “LNS” ( log network service).
- LNS sucks data either from Redo buffer cache (or) from online Redologs and shift’s the data to the physical standby
- Whenever there is an Archive log gap , RFS pings back the primary Archiver to shift the Archive log to the physical stand by.
- With the above Architecture ,we can minimize data mismatch, but we cant achicve “ A NO DATA LOSS SOLUTION”.
- In order to achive no data loss solution oracle as introduced stand by redo logs which needs to be created at the physical standby .
- The size of standby redo logs should be same as the size of primary redologs.
- In the above architecture , user gets commit complete acknowledgement , once the data as been returns to the primary redo logs as well as the stand by redo logs.
- If primary is unable to send the data to at least one stand by , primary goes down
- The no. of stand by redo logs that we need to create are
(max no. of redo logs per thread +1)*( max no. of thread)
(2+1)* 1 =3
- Thread is nothing but
- The above configuration mode is called maximum protection mode .
- With the above architecture , we can achive NO DATA LOSS SOLUTION .