Managing Backups and Recovery of the OCR Using OCR Backup Files

Two methods are most popular for copying OCR content and using it for recovery. The first method uses automatically generated OCR file copies and the second method uses manually created OCR export files.

Backing up OCR
For taking OCR backup we can use one of the following methods 
1.Backup from automatically generated OCR copies 
The Oracle Clusterware automatically creates OCR backups every four hours. At any one time, Oracle always retains the last three backup copies of the OCR. The CRSD process that creates the backups also creates and retains an OCR backup for each full day and at the end of each week.

Note:- You cannot customize the backup frequencies or the number of files that oracle retains. You can use any backup software to copy the automatically generated backup files at least once daily to a different device from where the primary OCR resides.

The default location for generating backups on UNIX-based systems is CRS_home/cdata/cluster_name where cluster_name is the name of your cluster. The Windows-based default location for generating backups uses the same path structure.
ocrconfig utility can be used to view the backups generated automatically by Oracle Clusterware.To find the most recent backup of the OCR, On any cluster node run following command
[root@rac2 ~]# ocrconfig -showbackup

rac2     2013/10/19 14:26:26     /app/gridHome/cdata/rac-cluster/backup00.ocr

rac2     2013/10/19 10:26:16     /app/gridHome/cdata/rac-cluster/backup01.ocr

rac2     2013/10/19 06:26:14     /app/gridHome/cdata/rac-cluster/backup02.ocr

rac2     2013/10/18 02:26:02     /app/gridHome/cdata/rac-cluster/day.ocr

rac2     2013/10/17 22:26:00     /app/gridHome/cdata/rac-cluster/week.ocr
PROT-25: Manual backups for the Oracle Cluster Registry are not available

If you are not getting any output , check the same command on other nodes 
To make a backup, use the ocrconfig tool with the showbackup option to identify the backup files, and then copy each file to a location that is redundant to the location of the original OCR backup file. This ensures that there are at least two copies of each OCR backup file.
Note:- You must be logged in as root user to run ocrconfig commands.if ocrconfig command is not working in root user, update the .bash_profile with specific bin path.

2.Manually creating OCR export files 
In addition to using the automatically created OCR backup files, you should also export the OCR contents to a file before and after making significant configuration changes, such as adding or deleting nodes from your environment, modifying Oracle Clusterware resources, or creating a database. Exporting the OCR contents to a file lets you import the OCR if your configuration changes cause errors. For example, if you have unresolvable configuration problems, or if you are unable to restart your cluster database after such changes, then you can restore your configuration by importing the saved OCR content from the valid configuration.
To perform manual backup of OCR:
You can  either use ocrconfig -manualbackup or ocrconfig -export command to force Oracle Clusterware to perform a backup of the OCR at any time, rather than wait for the automatic backup that occurs at 4-hour intervals.
The -manualbackup option is especially useful when you to need to obtain a binary backup on demand, such as before you make changes to the OCR. You need to be logged in as root user to take manual backups
To view the available OCR Manual backup
[root@rac1 ~]#ocrconfig -showbackup manual  
To perform manual backup
[root@rac1 ~]# ocrconfig -manualbackup
[root@rac1 ~]# ocrconfig -showbackup manual
rac1     2013/10/21 16:26:48     /app/gridHome/cdata/rac-cluster/backup_20131021_162648.ocr

You can also use export command along  with ocrconfig  to take the backup of  OCR  

[root@rac1 ~]# ocrconfig -export /backup/ocrbkp.dmp
[root@rac1 ~]# ll /backup/
-rw-r--r-- 1 root root 95055 Oct 26 17:52 ocrbkp.dmp

Recovering the OCR
If an application fails, try to restart the application and if you think the error is related to OCR, run an ocrcheck. It would normally return failure message if there is a issue or corruption in OCR. Then before attempting to restore the OCR, as a definitive verification that the OCR failed on both the primary OCR and the OCR mirror, Attempt to correct the problem using one of the following platform-specific OCR diagnostic procedures. If diagnosis doesn’t help you may have to restore OCR from backup
There are two methods for recovering the OCR. The first method uses automatically generated OCR file copies and the second method uses manually created OCR export files.

1.Restoring the OCR from Automatically Generated OCR Backups
Before you attempt to restore the OCR, ensure that the OCR is unavailable. To check the status of the OCR, Run the following command:
[root@rac1 ~]# ocrcheck
Status of Oracle Cluster Registry is as follows :
         Version                  :          3
         Total space (kbytes)     :     262120
         Used space (kbytes)      :       2696
         Available space (kbytes) :     259424
         ID                       : 1005378644
         Device/File Name         :       +OCR
                                    Device/File integrity check succeeded

                                    Device/File not configured

                                    Device/File not configured

                                    Device/File not configured

                                    Device/File not configured

         Cluster registry integrity check succeeded
         Logical corruption check succeeded
Note:-  Device/File not configured means ocrmirror is not configured
[root@rac1 ~]#
If both the primary OCR and the OCR mirror have failed. You must restore the OCR from a backup.If there is at least one copy of the OCR available, you can use that copy to restore the other copies of the OCR.When restoring the OCR from automatically generated backups, you first have to determine which backup file you will use for the recovery.
Log in as the root user and identify the available OCR backups using the ocrconfig command:
[root@rac1 ~]#ocrconfig -showbackup
Review the contents of the backup using the following ocrdump command, where file_name is the name of the OCR backup file:
[root@rac1 ~]# ocrdump -backupfile file_name
For recovering OCR from automated or manual backup needs all cluster, RAC instances and RAC database bring down before performing recovery of OCR.
[root@rac1 ~]# srvctl stop database -d MYRAC -o immediate
[root@rac1 ~]# srvctl status database -d MYRAC
Instance myrac1 is not running on node rac1
Instance myrac2 is not running on node rac2
[root@rac1 ~]# crsctl stop crs
Repeat this command on each node in your Oracle RAC cluster.
As root user , restore the OCR file using following command
[root@rac1 ~]#ocrconfig -restore file_name
As the root user, restart Oracle Clusterware on all the nodes in your cluster by restarting each node
[root@rac1 ~]# crsctl start crs

[root@rac1 ~]#srvctl start database -d MYRAC

Repeat this command on each node in your Oracle RAC cluster.
Use the Cluster Verification Utility (CVU) to verify the OCR integrity using following command. for that you have to login as grid user
[grid@rac1 ~]$  cluvfy comp ocr -n all 

2.Recovering the OCR from an OCR Export File
To restore the previous configuration stored in the OCR from an OCR export file:
Place the OCR export file that you created previously using the ocrconfig -export command in an accessible directory on disk.Stop Oracle Clusterware and RAC database on all the nodes in your Oracle RAC cluster as root user using following command.
[root@rac1 ~]# srvctl stop database -d MYRAC -o immediate
[root@rac1 ~]# srvctl status database -d MYRAC
Instance myrac1 is not running on node rac1
Instance myrac2 is not running on node rac2
[root@rac1 ~]# crsctl stop crs
Repeat this command on each node in your Oracle RAC cluster.
As root user restore the OCR data by data by importing the contents of the OCR export file
[root@rac1 ~]# ocrconfig -import file_name
Restart Oracle Clusterware on all the nodes in your cluster by restarting each node as root user using following command
[root@rac1 ~]# crsctl start crs

[root@rac1 ~]#srvctl start database -d MYRAC

Repeat this command on each node in your Oracle RAC cluster.
To verify the OCR integrity using Cluster Verification utility using following command.
[grid@rac1 ~]$ cluvfy comp ocr -n all 
 Want to know more about Real Application Cluster Click Oracle Real Application Cluster

Things to remember,

• Oracle takes physical backup of OCR automatically.
• No Cluster downtime or RAC database down time requires for PHYSICAL backup of OCR.
• No Cluster downtime or RAC database down time requires for MANUAL export backup of OCR.
• For recovery of OCR from any of above backup it should need to down ALL.
• All procedure requires ROOT login.
• As of 11gR2  when you backup OCR, it takes backup of voting disk automatically as per Oracle documentation.
• ocrconfig --manualbackup is introduced in 11g 
• The -showbackup option now includes auto and manual flags that you can optionally specify to view the auto/manual existing backup information.

For more information visit :- http://www.dbametrix.com/ocr-backup-recovery.html

Partitioning an existing table using datapump - a sample demo

First create a sample  table for testing purpose, 
SQL>CREATE TABLE big_table (
  id            NUMBER(10),
  created_date  DATE,
  lookup_id     NUMBER(10),
  data          VARCHAR2(50)
);

Execute the following procedure to insert large number of rows ,

DECLARE
  l_lookup_id    NUMBER(10);
  l_create_date  DATE;
BEGIN
  FOR i IN 1 .. 1000000 LOOP
    IF MOD(i, 3) = 0 THEN
      l_create_date := ADD_MONTHS(SYSDATE, -24);
      l_lookup_id   := 2;
    ELSIF MOD(i, 2) = 0 THEN
      l_create_date := ADD_MONTHS(SYSDATE, -12);
      l_lookup_id   := 1;
    ELSE
      l_create_date := SYSDATE;
      l_lookup_id   := 3;
    END IF;
    INSERT INTO big_table (id, created_date, lookup_id, data)
    VALUES (i, l_create_date, l_lookup_id, 'This is some data for ' || i);
  END LOOP;
  COMMIT;
END;
/
-- Apply some constraints and index to the table.

SQL> ALTER TABLE big_table ADD (CONSTRAINT big_table_pk PRIMARY KEY (id));

SQL>CREATE INDEX bita_created_date_i ON big_table(created_date);

Now big_table is created 

SQL>  select table_name, tablespace_name, partitioned
      from dba_tables where table_name like 'BIG_%';

TABLE_NAME                     TABLESPACE_NAME                PAR
------------------------------ ------------------------------ ---
BIG_TABLE                      EXAMPLE                        NO  ----->(No partition)

Export the table using datapump:
C:\Windows\system32>expdp admin/admin@oradb1 tables=sample.big_table directory=ar1 dumpfile=big_table.dmp parallel=4

Screen shot of my expdp:
--------------------------
Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - 64bit Production
With the Partitioning, Real Application Clusters, Automatic Storage Management, OLAP,
Data Mining and Real Application Testing options
Starting "ADMIN"."SYS_EXPORT_TABLE_01":  admin/********@oradb1 tables=sample.big_table directory=ar1 dumpfile=big_table.dmp parallel=4
Estimate in progress using BLOCKS method...
Processing object type TABLE_EXPORT/TABLE/TABLE_DATA
Total estimation using BLOCKS method: 55 MB
. . exported "SAMPLE"."BIG_TABLE"                        46.61 MB 1000000 rows
Processing object type TABLE_EXPORT/TABLE/TABLE
Processing object type TABLE_EXPORT/TABLE/INDEX/INDEX
Processing object type TABLE_EXPORT/TABLE/CONSTRAINT/CONSTRAINT
Processing object type TABLE_EXPORT/TABLE/INDEX/STATISTICS/INDEX_STATISTICS
Master table "ADMIN"."SYS_EXPORT_TABLE_01" successfully loaded/unloaded
******************************************************************************
Dump file set for ADMIN.SYS_EXPORT_TABLE_01 is:
  /u01/software/datapump/big_table.dmp
Job "ADMIN"."SYS_EXPORT_TABLE_01" successfully completed at 14:52:38

Now drop and recreate the table ,
SQL> drop table big_table;

Table dropped.

SQL> CREATE TABLE big_table (
    id            NUMBER(10),
    created_date  DATE,
    lookup_id     NUMBER(10),
    data          VARCHAR2(50)
)
PARTITION BY RANGE (created_date)
(PARTITION big_table_2003 VALUES LESS THAN (TO_DATE('01/01/2004', 'DD/MM/YYYY')),
 PARTITION big_table_2004 VALUES LESS THAN (TO_DATE('01/01/2005', 'DD/MM/YYYY')),
 PARTITION big_table_2005 VALUES LESS THAN (MAXVALUE));

SQL>  select table_name, tablespace_name, partitioned
      from dba_tables where table_name like 'BIG_%';

TABLE_NAME                     TABLESPACE_NAME                PAR
------------------------------ ------------------------------ ---
BIG_TABLE                      EXAMPLE                        YES ----> TABLE IS PARTITIONED

SQL> select count(*) from big_table;

  COUNT(*)
----------
         0  ----------> no data

Import the table using datapump
C:\Windows\system32> impdp admin/admin@oradb1 directory=ar1 dumpfile=big_table.dmp logfile=big_table.log table_exists_action=append parallel=4

note: job will be completed with one error because of "TABLE_EXISTS_ACTION=append" parameter.Use ignore=y if you are using old export utility.
Exclude Metadata and copy only data to avoid the error caused by the above parameter

screen shot:
-------------
Import: Release 11.2.0.1.0 - Production on Tue Jun 28 15:11:53 2011

Copyright (c) 1982, 2009, Oracle and/or its affiliates.  All rights reserved.

Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - 64bit Production
With the Partitioning, Real Application Clusters, Automatic Storage Management, OLAP,
Data Mining and Real Application Testing options
Master table "ADMIN"."SYS_IMPORT_FULL_01" successfully loaded/unloaded
Starting "ADMIN"."SYS_IMPORT_FULL_01":  admin/********@oradb1 directory=ar1 dumpfile=big_table.dmp logfile=big_table.log table_exists_action=append parallel=4
Processing object type TABLE_EXPORT/TABLE/TABLE
ORA-39152: Table "SAMPLE"."BIG_TABLE" exists. Data will be appended to existing table but all dependent metadata will be skipped due to table_exists_action of append
Processing object type TABLE_EXPORT/TABLE/TABLE_DATA
. . imported "SAMPLE"."BIG_TABLE"                        46.61 MB 1000000 rows
Processing object type TABLE_EXPORT/TABLE/INDEX/INDEX
Processing object type TABLE_EXPORT/TABLE/CONSTRAINT/CONSTRAINT
Processing object type TABLE_EXPORT/TABLE/INDEX/STATISTICS/INDEX_STATISTICS
Job "ADMIN"."SYS_IMPORT_FULL_01" completed with 1 error(s) at 15:12:49

SQL> select count(*) from big_table;

  COUNT(*)
----------
   1000000


SQL> insert into big_table (id, created_date, lookup_id, data)
  2  values(1001,'27-jun-03',5,'This is some old data');

1 row created.


SQL> select * from big_table partition(big_table_2004);

no rows selected

SQL> select * from big_table partition(big_table_2003);

        ID CREATED_D  LOOKUP_ID DATA
---------- --------- ---------- -------------------------
      1001 27-JUN-03          5 This is some old data

TESTING PARTITION PRUNING

SQL>  SET AUTOTRACE TRACEONLY EXPLAIN
SQL> select * from big_table
  2  where created_date='27-JUN-03';

Execution Plan
----------------------------------------------------------
Plan hash value: 1710340555

----------------------------------------------------------------------------------------
| Id  | Operation              | Name      | Rows  | Bytes | Cost (%CPU)| Time     | Pstart| Pstop |
----------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT       |           |     1 |    37 |     3   (0)| 00:00:01 |       |       |
|   1 |  PARTITION RANGE SINGLE|           |     1 |    37 |     3   (0)| 00:00:01 |   KEY |   KEY |
|*  2 |   TABLE ACCESS FULL    | BIG_TABLE |     1 |    37 |     3   (0)| 00:00:01 |   KEY |   KEY |
----------------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

   2 - filter("CREATED_DATE"='27-JUN-03')

Source:- http://arundaskalathil.blogspot.in/2011/06/partitioning-existing-table-using.html

Oracle Cluster tables - an efficient way to reduce the I/O

If you two are more tables are joined together on a single column and most of the time you issue join queries on them, then consider creating a cluster of these tables.
 A cluster is a group tables that share the same data blocks i.e. all the tables are physically stored together.

For example EMP and DEPT table are joined on DEPTNO column. If you cluster them, Oracle physically stores all rows for each department from both the emp and dept tables in the same data blocks.
Since cluster stores related rows of different tables in same data blocks, Disk I/O is reduced and access time improves for joins of clustered tables.
Each cluster key value is stored only once each in the cluster and the cluster index, no matter how many rows of different tables contain the value.
Therefore, less storage might be required to store related table and index data in a cluster than is necessary in non-clustered table format.

Creating a Clustered Table 
 To create clustered tables. First, create a cluster and create index on it. Then create tables in it.
For example to create a cluster of EMP and DEPT tables in which the DEPTNO will be cluster key, first create the cluster by typing the following command.

SQL>create cluster emp_dept (deptno number(2));

Then create index on it.

SQL>create index on cluster emp_dept;

Now create table in the cluster like this

SQL>create table dept (deptno number(2),
                    name varchar2(20),
                    loc varchar2(20))
                    cluster emp_dept (deptno);

SQL>create table emp (empno number(5),
            name varchar2(20),
            sal number(10,2),
            deptno number(2)) cluster emp_dept (deptno)      

Dropping Clusters
To drop a cluster use DROP CLUSTER statement. For example to drop the emp_dept cluster give the following command.

SQL>drop cluster emp_dept;

This will drop the cluster, if the cluster is empty i.e. no tables are existing it it. If tables are there in the cluster first drop the tables and then drop the cluster. If you want to drop the cluster even when tables are there then give the following command.

SQL>drop cluster emp_dept including tables;

Listing Information about Clusters
To see how many clusters are there in your schema give the following statement.

SQL>select * from user_clusters;

To see which tables are part of a cluster. Give the following command.

SQL>select * from tab

TABLE_NAME    TYPE            CLUSTER_ID
----------             ----              -----------
EMP                  TABLE         1
SALGRADE        TABLE        
CUSTOMER        TABLE        
DEPT                TABLE         1
In the above example notice the CLUSTER_ID column, for EMP and DEPT table the cluster_id is 1. That means these tables are in cluster whose cluster_id is 1. You can see the cluster_id’s name in USER_CLUSTERS table.
Source:- http://www.oracle-dba-online.com/

Installing Oracle 10g RAC on Oracle VM in windows 2008 using openfiler as shared storage

Dear friends those who interested in knowing more about RAC architecture , they need to prepare one RAC system for their study purpose.Here i got some helpful information from you-tube that provides detailed information  about installing Oracle 10g RAC on windows  .
Thanks Mr.Ahmed D.sherif for his nice presentation.  Here in his demonstration he uses Oracle VM as virtual box, windows server 2003 (32bit) as guest operating system , openfiler+Microsoft iscsi  initiator as shared storage ,oracle database and clusterware (10.2.0.1).

But some softwares listed above  are not available for me and i completed the same setup using  windows server 2008 standard edition(64bit) and oracle software (10.2.0.4) (both clusterware and database software).
The youtube link that i followed is given below,
Part1
http://www.youtube.com/watch?v=6BmwrR3Vcpc
Part2
http://www.youtube.com/watch?v=w7foLjbac4c
Part3
http://www.youtube.com/watch?v=BMPd_XAmtdc
Part4
http://www.youtube.com/watch?v=shVTEpyBohI
Part5
http://www.youtube.com/watch?v=shVTEpyBohI

You can fulfill the RAC set up by following  above link . But i hitched in several stage ,and it eat up my valuable time . Here i want to share that experience.

1. You can install Openfiler software by watching the video , during installation you will prompted for root user password. After installation you will get an url for openfiler interface just like
https://192.168.1.141:446 , it will ask for username and password
Here you have to give  'openfiler' as username not 'root'  ,as  i don't know the default password for 'openfiler' user , so i reset the password by 'root' user.

2.You don't need to download microsoft iscsi software  for windows 2008 server , as it comes in built with windows 2008 server  editions.

3. Once you configured shared disk for RAC in openfiler , You  have to configure the disk on each nodes through microsoft-iscsi. After configuring microsoft-iscsi you can see the shard disk has been appeared in 'Disk Management'  window on every node. Each disk should be in 'online' mode.After that you have to partition each disk (here 5) without any file system and drive name. After these stage i struggled a lot , because while cluster-ware installation it did not detect my previously created shared disk.After googling and studying a lot of articles for more than one weeks, i came to know that in windows clusterware does not support partition created as primary .In windows 2008 server ,by default  it creates partition as 'primary' partition .So you have to use diskpart command to create extended partition in windows 2008. So i deleted every partition that i mounted from storage. Then i created extended partition by using diskpart command.
How to use diskpart command 
1. Take command prompt and run diskpart command 
then type 
DISKPART>list disk
it will show all available disks, 
2.Then select the disk that you want to make extended partition by
DISKPART>select disk 1 
3.Then create extended partition on the selected disk  
DISKPART> create partition extended
You have to follow the same step for remaining disk . Once you created extended partition for all allocated disk , you can create logical partition by right clicking  each disk in the  'Disk Management' window.While creating logical drive keep in mind that not to format the disk  drive also not assign any drive letter.  The end result  become 5 logical disk (as per the video) and are ready to recognized by the clusterware software. Once the disks are ready 
you can go ahead with clusterware installation .

Note:- 
In oracle 10g raw devices are used for storing voting disk and ocr , but from 11g onwards you can keep these things in ASM . So here first two raw device are used for voting disk and ocr (both are configured in external redundancy - having only one disk) and remaining three raw devices are used for ASM.
Rest of my installation never stuck in any stage and i competed my two node RAC setup.

Hope it helps somebody those interested in oracle RAC architecture. 
Enjoy the reading :)
  

Different types of Failover mechanism in Oracle

FAILOVER: 
In the context of Oracle Net, failover refers to the mechanism of switching over to an alternate resource when connection to the primary resource gets terminated due to any reason. Connection failure may be broadly categorized as:
Those that occur while making the initial connection. (Connect Time Failover)
Those that occur after a connection has been successfully established. (TAF)

The first category of connection failure can be dealt with comparatively easily. If your attempt to connect to an instance fails, you can make the connection attempt again, but this time to a backup instance. As long as you have backup instances configured, you can continue trying to connect, until a successful connection is established. This process is technically termed as Connect Time Failover.

Connect Time Failover.
The connect time failover feature allows client to connet to another listener if the initial connection to the first listener fails.Multiple listener locations are specified in the client tnsnames.ora file. If a connection attempt to the first listener fails , a connection request to the next listener in the list is attempted .You can acheive this feature by adding Multiple listener addresses within an address list of tnsnames.Here is an example of what a tnsnames.ora file looks like with a connect-time failover enabled.

TNS Parameter for Connect time failover is FAILOVER. Default value of this is ON.
(failover=on) is default for ADDRESS_LISTs and DESCRIPTION_LISTs , so we do not need to explicitly specify this parameter to put it on.

Sample for TNS net service alias for Client Side Connect time Failover :-

myrac_failover =  
(DESCRIPTION =  
 (ADDRESS_LIST =  
 (ADDRESS = (PROTOCOL = TCP)(HOST = rac1-vip)(PORT = 1522))
  (ADDRESS = (PROTOCOL = TCP)(HOST = rac2-vip)(PORT = 1522))  
 )  
 (CONNECT_DATA =  
 (SERVICE_NAME = myrac)  
 )  
 )

Notice the additional entry under the ADDRESS_LIST section. Here two listener are specified .If a connection is unsuccessful  when attempting to connect to the rac1-vip host on port 1521, a connection attempt is made to the rac2-vip host on port 1521 . 

Note:- Entries in the ADDRESS_LIST do not have to be a RAC nodes . The example that i explained here is for two node RAC database. You can also configure connect-time failover with a standby database , so that one of the entries in the list may be a standby database .As long as the defined service(Database) in the ADDRESS_LIST is availabe on the other 
node , client is able to connect to that node.
For example,

failover_norac=
(DESCRIPTION=
(ADDRESS_LIST=
(ADDRESS= (PROTOCOL=TCP) (HOST=prod_db) (PORT=1521))
(ADDRESS= (PROTOCOL=TCP) (HOST=standby_db) (PORT=1521))
(FAILOVER= TRUE)
)
(CONNECT_DATA=
(SERVICE_NAME= orcl)
)
)
The other criteria, of cource , that the address in the list will allow the client to get to the data that is needed by the application. With oracle RAC , we know that data is always the same, as it is the same database. In the case of physical or logical standby database  
whether or not the data is accessible depends upon how these standby database is configured.

Note:- An important issue to be aware of at this point is that Connect Time Failover only works for you, if you are using dynamic registration. 

Transparent application failover (TAF)
Now, let’s look at how TAF works. Unlike connect time failover, which is invoked before the connection is made, TAF comes into play after the connection is made (and then, broken). If the connection is lost while the application is running, Oracle Net will transparently reconnect the application to another instance accessing the same database.

TAF supports two types of failover: SESSION and SELECT. A SESSION failover connection is not over ambitious. It just fails over to a backup instance. All work in progress at that point are irrevocably lost. SELECT failover is more intricate in as much as it enables some type of read only application to failover without losing the work in progress. If a SELECT statement was in progress at the time of the termination of the connection, then as the connection is reestablished to a backup instance, Oracle Net re-executes the SELECT statement and positions the cursor in a manner that the client can seamlessly continue fetching the rows. But that’s about all that TAF has to offer. It doesn’t have any mechanism to recover DML statements that were in progress, when the failure occurred, or even for SELECT statements, you lose global temporary tables, package states and session settings.
TAF supports two failover methods: BASIC and PRECONNECT. In BASIC method, you connect to the backup instance when the primary connection fails. In the PRECONNECT method, you connect to the backup instance at the same time you connect to the primary instance. This has the obvious benefit of having a backup connection available all of the time, thus reducing the time of ‘failover’. But the downside is that you have to pay the extra ‘cost’ in terms of resources spent, of having a backup connection open all the while.
TAF is configured by adding a FAILOVER_MODE parameter to the CONNECT_DATA parameter for a net service name. Below is an example for configuring TAF with two node RAC

Note:-
Specifies a different net service name to be used to establish the backup connection. A backup should be specified when using PRECONNECT to pre-establish connections. Specifying a BACKUP is strongly recommended for BASIC methods; otherwise, reconnection might first attempt the instance that has just failed, adding additional delay until the client reconnects. 
http://oracleinquisitive.blogspot.in/2012/09/transparent-application-failover-taf.html

How load balancing works in oracle ?

Load Balancing
Oracle RAC systems provide two types of load balancing for automatic workload management:
* Server load balancing distributes processing workload among Oracle RAC nodes.
* Client load balancing distributes new connections among Oracle RAC nodes so that no one server is overwhelmed with connection requests. For example, when a connection fails over to another node because of hardware failure, client load balancing ensures that the redirected connection requests are distributed among the other nodes in the RAC.
The primary difference between these two methods is that the former method distributes processing and the latter method distributes connection attempts.Lets look in details

Client side load balancing 
Client side load balancing is determined by configuring the tnsnames.ora with multiple address in the description ,with parameter LOAD_BALANCE=yes (can specify 'ON' also) , as in this example .
myrac =
  (DESCRIPTION =
   (LOAD_BALANCE=ON)
   (ADDRESS = (PROTOCOL = TCP)(HOST = rac1-vip)(PORT = 1521))
   (ADDRESS = (PROTOCOL = TCP)(HOST = rac2-vip)(PORT = 1521))
    (CONNECT_DATA =
     (SERVICE_NAME = myrac)
    )
  )

Client load balancing helps distribute new connections in your environment so that no one server is overwhelmed with connection requests. When client load balancing is enabled, connection attempts are made randomly among RAC nodes. Suppose you have the Oracle RAC environment  with multiple Oracle RAC nodes, A, B, C, and D. Without client load balancing enabled, connection attempts may be front-loaded, meaning that most connection attempts would try Node A first, then Node B, and so on until a connection attempt is successful.This creates a situation where Node A and Node B can become overloaded with connection requests. Thus in client side load balancing the client will randomly pick an address to try ,without regards to the availability of the node or to how heavily loaded the node may be.

Server Load Balancing
Server side load balancing is controlled by two parameters LOCAL_LISTENER and REMOTE_LISTENER.By default LOCAL_LISTENER is set to blank . Setting local_listener parameter is mandatory when using a different port rather than default .LOCAL_LISTENER on each node should point to the listener on that node.Oracle database have the ability to automatically register their 
presence with an existing default listener with the help of pmon , or you must specify the init parameter LOCAL_LISTENER for non default listener .

A remote listener is a listener residing on one computer that redirects connections to a database instance on another computer. Remote listeners are typically used in an Oracle Real Application Clusters (Oracle RAC) environment. You can configure registration to remote listeners, such as in the case of Oracle RAC, for dedicated server or shared server environments. The REMOTE_LISTENER parameter in the spfile should be pointing to a TNSNAMES entry on each server mode, which in turn list the IPs and ports of all available listeners in the cluster. DBCA will set these entries up for you if you use it to create your databases (recommended).

The remote_listener parameter is then set to listeners_myrac in the spfile

*.remote_listener='LISTENERS_MYRAC'

In tnsnames.ora you can see entry like this,

LISTENERS_MYRAC =
  (ADDRESS_LIST =
    (ADDRESS = (PROTOCOL = TCP)(HOST = rac1-vip )(PORT = 1521))
    (ADDRESS = (PROTOCOL = TCP)(HOST = rac2-vip )(PORT = 1521))
  )

There are two types of server-side load balancing: 
Load Based — Server side load balancing redirects connections by default depending on node load. This id default.
Session Based — Session based load balancing takes into account the number of sessions connected to each node and then distributes the connections to balance the number of sessions across the different nodes.

Note:- Use listener_<instance_name> as the alias name for the local listener, and listeners_<servicename> for the remote listener alias.
The way you set the remote listener changed in rac 11.2 , instead of registering with local listener, the database register with SCAN listener.
This approach is similar to above, but the pain of configuring individual nodes/their VIPs in client tnsnames.ora is eliminated. All we need is a Round Robin enabled DNS name resolution and use the SCAN DNS name in the client tnsnames.ora.If you do not use DBCA, you should set the REMOTE_LISTENER database parameter to scan_name:scan_port. 

Server-side load balancing divides the connection load evenly between all available listeners by determining the total number of connections on each listener, and then distributing new user session connection requests to the least loaded listener(s) based on the total number of sessions already connected. While a bit more complex to implement because it requires configuration of multiple listeners, it most definitely helps to even out connections across all available listeners in a database system.

To implement server-side load balancing, at least two listeners must be configured. Also, the REMOTE_LISTENERS initialization parameter must be added to the database’s PFILE or SPFILE so that the database knows to search out the value provided in that parameter in the database server’s TNSNAMES.ORA configuration file. When server-side load balancing is activated, each listener that contributes a listening endpoint communicates with the other listener(s) via each database instance’s PMON process. Oracle then determines how many user connections each listener is servicing, and it will distribute any new connection requests so that the load is balanced evenly across all servers. The entries in TNSNAMES.ORA direct the listeners to share information about the relative load connectivity.
   

Data Guard Protection Modes and redo transport modes in oracle

Log Transport Services enables and controls the automated transfer of redo data within a Data Guard configuration from the primary site to each of its standby sites.
Log transport services also controls the level of data protection for your database. The DBA will configure log transport services to balance data protection and availability against database performance. Log transport services will also coordinate with Log Apply Services and Role Transition Services for switchover and failover operations.

Maximum Performance Mode
Maximum Performance mode is the default protection mode and provides the highest level of data protection that is possible without affecting the performance or availability of the primary database. With this protection mode, a transaction is committed as soon as the redo data needed to recover the transaction is written to the local (online) redo log.
When configuring the standby destination service in the LOG_ARCHIVE_DEST_n initialization parameter on the primary database, log transport services can be set to use either LGWR / ASYNC or ARCH. In order to reduce the amount of data loss on the standby destination if the primary database were to fail, set the LGWR and ASYNC attribute. Using this configuration, the primary database writes its redo stream to the standby redo logs on the standby database asynchronously with respect to the commitment of the transactions that create the redo data. When the nodes are connected with sufficient bandwidth, this mode provides a level of data protection that approaches that of Maximum Availability mode with minimal impact on primary database performance.
Note that the use of standby redo logs while operating in Maximum Performance mode is only necessary when configuring log transport services to use LGWR. When log transport services is configured to use ARCH, standby redo logs are not required.
The minimum requirements are described in the following table:

For example,
log_archive_dest_2='service=testdb_standby ARCH NOAFFIRM'
or
log_archive_dest_2='service=testdb_standby LGWR ASYNC NOAFFIRM'

After modifying these setting in init file ,execute the following command from sql prompt 
SQL>  ALTER DATABASE SET STANDBY DATABASE TO MAXIMIZE PERFORMANCE  ;
The above command is actually not required as MAXIMIZE PERFORMANCE is the default redo transport mode in data guard .

Maximum Protection Mode
Maximum Protection mode offers the ultimate in data protection. It guarantees no data loss will occur in the event the primary database fails. In order to provide this level of protection, the redo data needed to recover each transaction must be written to both the local (online) redo log and to a standby redo log on at least one standby database before the transaction can be committed. In order to guarantee no loss of data can occur, the primary database will shut down if a fault prevents it from writing its redo data to at least one remote standby redo log.
In a multiple-instance RAC database environment, Data Guard will shut down the primary database if it is unable to write the redo data to at least one properly configured database instance (see minimum requirements below).
In order to participate in Maximum Protection mode:
At least one standby instance has to be configured with standby redo logs.
When configuring the standby destination service in the LOG_ARCHIVE_DEST_n initialization parameter on the primary database, you must use the LGWR, SYNC, and AFFIRM attributes.
The minimum requirements are described in the following table:

For example:
log_archive_dest_2='service=testdb_standby LGWR SYNC AFFIRM'
and
SQL>  ALTER DATABASE SET STANDBY DATABASE TO MAXIMIZE PROTECTION   ;
will keep the data guard in maximum protection mode . 
The SYNC attribute with the LGWR process specifies that network I/O is to be performed synchronously for the destination. This means that once the I/O (redo data) is initiated, the archiving process waits for the I/O to complete before continuing. When specifying the SYNC attribute, all network I/O operations are performed synchronously, in conjunction with each write operation to the online redo log. The transaction is not committed on the primary database until the redo data necessary to recover that transaction is received by the destination.

It is highly recommended that a Data Guard configuration operating in Maximum Protection mode contain at least two physical standby databases that meet the requirements listed in the table above. That way, the primary database can continue processing if one of the physical standby databases cannot receive redo data from the primary database. If only one standby database is configured with the minimum requirements listed above, the primary database will shut down when the physical standby databases cannot receive redo data from the primary database!

Net Timeout
The Data Guard environment works by sending the redo data to the standby server by connecting to the database instance there. If the instance does not respond in time, the log shipping service will wait for a specified timeout value and then give up. This timeout value can be set in Oracle Database, using a parameter called net_timeout. In maximum protection mode, the log shipping service will retry for 20 times before giving up.


Maximum Availability Mode
Maximum Availability mode provides the highest level of data protection that is possible without affecting the availability of the primary database. This protection mode is very similar to Maximum Protection where a transaction will not commit until the redo data needed to recover that transaction is written to both the local (online) redo log and to at least one remote standby redo log. Unlike Maximum Protection mode; however, the primary database will not shut down if a fault prevents it from writing its redo data to a remote standby redo log. Instead, the primary database will operate in Maximum Performance mode until the fault is corrected and all log gaps have been resolved. After all log gaps have been resolved, the primary database automatically resumes operating in Maximum Availability mode.

Please note that Maximum Availability mode guarantees that no data will be lost if the primary fails, but only if a second fault does not prevent a complete set of redo data from being sent from the primary database to at least one standby database.
Just like Maximum Protection mode, Maximum Availability requires:
At least one standby instance has to be configured with standby redo logs.
When configuring the standby destination service in the LOG_ARCHIVE_DEST_n initialization parameter on the primary database, you must use the LGWR, SYNC, and AFFIRM attributes.
The minimum requirements are described in the following table:

For example:
log_archive_dest_2='service=testdb_standby LGWR SYNC AFFIRM'
and
SQL>  ALTER DATABASE SET STANDBY DATABASE TO MAXIMIZE  AVAILABILITY ;
will keep the data guard in maximum protection mode . 

Oracle Data Guard : Synchronous vs. Asynchronous Redo TransportData Guard Redo Transport Services coordinate the transmission of redo from a primary database

to the standby database. While LGWR process in Primary database is writing redo to its Online Redo Log files (ORL), a separate Data Guard process called the Log Network Server (LNS) is reading from the redo buffer in SGA and passes redo to Oracle Net Services for transmission to the standby database. Redo records transmitted by the LNS are received at the standby database by another Data Guard process called the Remote File Server (RFS) that writes it to a sequential file called a standby redo log file (SRL).

Synchronous Redo Transport
Its also called “zero data loss” method as the LGWR is not allowed to acknowledge a commit has succeeded until the LNS confirms that the redo needed to recover the transaction has been written to disk at the standby site.

So that's how it works

1. when user performs commits. The LGWR reads the redo record from the log buffer, writes it to the online redo log file, and waits for confirmation from the LNS.

2. The LNS reads the same redo record from the log buffer and transmits it to the standby database using Oracle Net Services. The RFS receives the redo at the standby database and writes it to a standby redo log file.

3. When the RFS receives a write-complete from the disk, it transmits an acknowledgment back to the LNS process on the primary database, which in turn notifies the LGWR that transmission is complete. The LGWR then sends a commit acknowledgment to the user.

This transport mode is used by the Maximum Protection and Maximum Availability data protection modes

Asynchronous Redo Transport

Asynchronous transport (ASYNC) LGWR  process does not wait for the acknowledgment from the LNS. This creates a near zero performance impact on the primary database regardless of the distance between primary and standby locations . The asynchronous redo transport mode transmits redo data asynchronously with respect to transaction commitment. A transaction can commit without waiting for the redo generated by that transaction to be successfully sent to any redo transport destination that uses the asynchronous redo transport mode. This transport mode is used by the Maximum Performance data protection mode