Data Guard 19c Data Guard 19c Data Guard 19c

Author’s name
Data Guard 19c
Carlos López
Advanced Services Engineer
Advanced Customer Services
16 de Marzo de 2022

Safe harbor statement
The following is intended to outline our general product direction. It is intended for
information purposes only and may not be incorporated into any contract. It is not
a commitment to deliver any material, code, or functionality, and should not be relied
upon in making purchasing decisions.
The development, release, timing, and pricing of any features or functionality described
for Oracle’s products may change and remains at the sole discretion of Oracle
Corporation.
2 Copyright © 2020, Oracle | Confidential: Public

Introduction to Data Guard
1
2
3
4
Concepts
Types of Standbys
Protection Modes
Role transitions
Copyright © 2022 and/or its affiliates. All rights reserved.Confidential | Oracle Public
3
5
Agenda
6 Configuration
7 Administration
8 Active Data Guard
9 RAC vs. DG
10 Scenarios

Introduction to DataGuard
1
2
3
4
Concepts
Types of Standbys
Protection Modes
Role transitions
4
5
Agenda
6 Configuration
7 Administration
8 Active Data Guard
9 RAC vs. DG
10 Scenarios

5
• Cases:
• Massive update in the DATABASE. Be able to try it before in snapshot stby
• Deleting a datafile by hand outside the database.
• Corruption of the BD. Problems in the cabin.
• Failure of the RAC that causes the databases to fall and they cannot be lifted.
• Test an upgrade before doing so in PRO
• BD production very loaded with work. Offloading
• ...
•

1
2
3
4
Concepts
Types of Standbys
Protection Modes
Role transitions
6
5
Agenda
6 Configuration
7 Administration
8 Active Data Guard
9 RAC vs. DG
10 Scenarios

7
Concepts.
Dataguard 1/2
Functionality available in Oracle Database Enterprise Edition (EE) since Oracle Database
version 9i.

8
Replication
It consists of anActive-Passive environment
consisting of our production database and one
or more standby databases. In the 19c version
there is the possibility to configure up to 30
standby.
The DB REDO is replicated
Alta disponibilidad
Protección y recuperación de datos
ante desastres (HA+DR).
Reuse of resources
Use of standby to free up resources used in
production, such as heavy procedures, backup,
etc.
Downtime
Reduction of downtime that is invested
in maintenance tasks, updates (Rolling
upgrades), etc.
Concepts.
Dataguard 2/2
100%

9
Concepts.
Requirements 1/2
• Versión Oracle Enterprise Edition
• Primary and standby environments can have different:
• Resources ( CPU, RAM)
• Storage (ASM or Filesystem)
• Architecture ( RAC or Single Instance)
• Both environments should have the same platform *
• Both environments should have the same OS version*
• Both environments should have the same RDBMS version*
• There are a few exceptions:
• Data Guard Support for Heterogeneous Primary and Physical Standbys in Same Data Guard Configuration [ID 413484.1]
• Data Guard Support for Heterogeneous Primary and Logical Standbys in Same Data Guard Configuration [ID 1085687.1]

• Different DB_UNIQUE_NAME
• Same value of the COMPATIBLE parameter
• ARCHIVELOG Mode
• It is recommended to configure FORCE LOGGING *
• It is not mandatory but prevents certain data from not being replicated and causing errors in the stby
•
* The Gains and Pains of Nologging Operations in a Data Guard Environment (Doc ID 290161.1)
10
Concepts.
Requirements 2/2

Concepts.
Basic architecture 1/3
• 1 BD Primary and only 1
• 1 or more STANDBYS
• Standby can be in:
• Same server
• On another server
• In another CPD
• Or a mix of the previous 3
• Redo transport services
• Replica services at the standby
•
11

12
Concepts.
DG 2/3 Architecture
Archived
Redo Logs
Online
Redo Logs
FAL
Oracle
Net
Primary
Database
Transactions
Physical/Logical
Standby
Database
Backup
/
Reports
LGWR RFS
Standby
Redo
Logs
Archived
Redo Logs
ARCH
MRP/ LSP
Transform Redo
to SQL for
SQL Apply
ARCH
NSS (SYNC)
NSA ( ASYNC)
sync
async
arch

13
Redo transport service
( NSS-NSA / RFS / ARCH )
• Controls the automatic sending of the redo
from the production database to the existing
standbys.
• Depending on the desired configuration mode
we will use one parameterization or another.
•
Redo Application Services
( MRP / LSP )
Applies the audio information in the standby to
maintain synchronization with the primary
MRP: for Physical standby
LSP: for Logical standby
The information is applied directly from the
standby redo logs if we use Real Time Apply.
Concepts.
Services 3/3

14
• Modos
• Maximum Performance
• (Maximum Performance):
• Async
• Default
• Maximum Protection
• (Maximum Protection):
• Sync
• Maximum Availability
• (Maximum Availability):
• Sync->Async
•
• Which is right for our environment?
• Business Needs
• Data loss
• Availability
• Yield
• Resources
• LAN / WAN - Bandwidth
•
Concepts.
Protection modes

15
• SQLPLUS • Data Guard BROKER (DGB)
- DGMGRL command-line interface
- Enterprise Manager Cloud Control
Concepts.
Management interfaces

1
2
3
4
Concepts
Types of Standbys
Protection Modes
Role transitions
16
5
Agenda
6 Configuration
7 Administration
8 Active Data Guard
9 RAC vs. DG
10 Scenarios

17
Types of BD Standby
• Physical
• Snapshot
• Active Data Guard
• Cascaded
• Far Sync
• Logical

18
Types of Standbys
Physical Standby
 It allows you to have an identical copy, block by block, of our production database.
The structure, schemes or indexes are the same.
 It is in MOUNT or OPEN in READ-ONLY
 Synchronizes using Redo Apply received from the primary (Media Recovery similar
to what RMAN recover does)

Use:
Fault data protection
Reporting, RMAN offloading

19
Types of Standby
Physical Standby. Snapshot Standby
La standby original tiene que ser de tipo Physical Standby
Con este procedimiento dispondremos puntualmente de nuestra base de datos standby en
modo READ-WRITE.
Como la snapshot standby sigue siendo un tipo de standby, va a seguir recibiendo
correctamente los archives aunque no los aplicara hasta que no vuelva a ser standby física.
Puede volver atrás y ser una Physical Standby normal
 Using Snapshot Standby Database. [Doc ID 443720.1]
– Use:
– Validations
– Reporting

20
Types of Standby
Physical Standby. Active Data Guard (ADG)
• The Active Dataguard (ADG) option is available starting with version 11.1 and requires
additional licensing.
• The physical standby is in a READ-ONLY state and, at the same time, applying changes.
This allows us to have a synchronized environment where to run ONLINE reporting.
• This functionality is also called Real-time Query
• The COMPATIBLE must be set at least to 11.0.0 in both the primary and standby
•
https://www.oracle.com/technetwork/database/availability/dg-adg-technical-overview-wp-
5347548.pdf

21
Tipos deStandbyTypes of StandbyPhysicalStandby.CascadedStandby
 The BD standby receives the redo from ANOTHER standby, instead of from the primary
 With cascading, the overload associated with sending the redo is performed by a BD PHYSICAL
standby cascading, rather than the BD primary.
 Only physical standbys can cascade redo
 A "cascading standby DB" can send the redo, either in real-time mode (as it is written to its standby
redo log files) or in non-real-time (when a standby redo log file has been completed and is being
archived in the "cascading standby" => lag secured)
 "Real-time cascading" requires an Oracle Active Data Guard (ADG) license
 Cascaded Standby Databases in Oracle 12c (Doc ID 1542969.1)
Use:
 To porteger locally and regionally
 If you have several standbys, so the primary does not
 you have to send the redo to all of them


22
Types of Standby
PhysicalStandby. Far SyncStandby
 Data Guard 12c New Feature: Far Sync Standby (Doc ID 1565071.1)
 New configuration available for standby. It works as a repository of archives.
(Proxy).
 The Standby Far Sync cannot take the role of primary, it only has control files,
standby redo logs and archives.
 It is in MOUNT mode
 Requires ADG license
 SYNC from primary to FSS.
 ASYNC from FSS to Stby => lag

Use:
 Cascading data guard
 Long distance data guard


23
Types of Standby
Logical Standby 1/2
 It contains the same logical information as the production database, although the
physical organization and structure may be different.
 You can add more objects to it than the Primary already has or even stop
synchronizing certain objects in it. This is configurable. By default you do not let
anything of the LS be modified
 It is OPEN in READ WRITE
 Stays in sync using SQL Apply
 Reconstructs SQL queries from the data received from the primary (LOGMINER) and
executes them in the standby.

Use:
 Fault data protection*
 Reporting
 Upgrades with minimal downtime


24
Prerequisites. Before converting a physical standby to logical we must:
Check for objects and data types that are not supported. If they are not, SQL Apply will simply ignore them
and not apply their changes.
Make sure the records are unique, otherwise replication in the stby will be slow
Know which commands are not replicated in the Logical Standby and must be executed manually in the stby
Types of Standby
Logical Standby 2/2
SQL> SELECT OWNER, TABLE_NAME
FROM DBA_LOGSTDBY_NOT_UNIQUE
WHERE (OWNER, TABLE_NAME) NOT IN
(SELECT DISTINCT OWNER, TABLE_NAME
FROM DBA_LOGSTDBY_UNSUPPORTED)
AND BAD_COLUMN = 'Y';
no rows selected
SQL> SELECT
OWNER,TABLE_NAME,COLUMN_NAME,DATA_TYPE
FROM DBA_LOGSTDBY_UNSUPPORTED ORDER BY
OWNER,TABLE_NAME,DATA_TYPE;
no rows selected

1
2
3
4
Concepts
Types of Standbys
Protection Modes
Role transitions
25
5
Agenda
6 Configuration
7 Administration
8 Active Data Guard
9 RAC vs. DG
10 Scenarios

26
Protection Modes
 Maximum Performance
 It is the default mode.
 It provides the maximum possible data protection without penalizing the performance of the
primary.
 Asynchronous
 Maximum Protection
 It ensures 100% that no data will be lost. Synchronous mode.
 A transaction will not be committed in the primary until that transaction is committed in at
least one of the standbys.
 The primary database may come to a standstill to prevent data loss.
 Maximum Availability
 Try to work in maximum protection and if an error occurs it goes to work as maximum performance.
 The time in which you try to maintain synchronous communication is limited by the parameter
NET_TIMEOUT.
 Synchronous and Asynchronous* (*only in case of transmission failure)

27
Protection Modes
Network and Write to Disk
Protection Mode Risk of Data Loss Requirements Redo Shipping
Maximum Performance
Minimal data loss – from
0 to a few seconds
ARC (Prior 11g)
LGWR ASYNC NOAFFIRM
Standby Redo Logs
Asynchronous
Maximum Availability
Zero Data Loss
Single Failure Protection
LGWR SYNC AFFIRM
Standby Redo Logs
Synchronous
Maximum Protection Zero Data Loss
Double Failure Protection
LGWR SYNC AFFIRM
Standby Redo Logs
Synchronous
LGWR SYNC NOAFFIRM
Transmission over network : SYNC / ASYNC
Disk Write Confirmation: AFFIRM / NOAFFIRM
By default when we have a SYNC synchronization we have a AFFIRM confirmation of redo
transport. And when it's ASYNC it will be NOAFFIRM.
FAST-SYNC= SYNC NOAFFIRM

Protection Modes
Real-timeApply
From 11gR2
The stby applies the redo as it arrives. He
doesn't wait for a stby redo log to fill out to
do so.
Therefore, there is less lag.
RFS – Remote File Server process. Receive
information from primary school redos
MRP – Managed Recovery Process. Apply the
information of the redos in the standby
LSP – Logical Standby Process.
Real-time apply requires the standby database to be configured with stanby redo logs and in archived
mode.
28

Protection Modes
FAST-SYNC
Primary Database Standby
Database
User Transactions
Queries, DML, DDL
Standby
Redo
Logs
RFS
NSS
Primary
Online
Redo Logs
LGWR
MRP
SGA
Redo
Buffer
Oracle Net
Commit
Data Guard Processes
NSS – transmits redo from the buffer log of the primary
RFS – receives redo from the primary and writes it to
the stby redo log.
But give the OK BEFORE writing it (green arrows)
MRP – Recovery process in the db stby
Synchronous without waiting for I/O writing at standby- LGWR SYNC NOAFFIRM
Copyright © 2022 and/or its affiliates. All rights reserved.
29 Copyright © 2022 and/or its affiliates. All rights reserved.Confidential | Oracle Public
29

1
2
3
4
Concepts
Types of Standbys
Protection Modes
Role transitions
30
5
Agenda
6 Configuration
7 Administration
8 Active Data Guard
9 RAC vs. DG
10 Scenarios

31
Role transitions
In a dataguard the following roles are available:
• Primary
• Standby
• Physics/Logic
• Snapshot StandBy
• Cascaded/Cascading Standby
• FarSync
Only one primary can exist.
Advantages
• Reduced downtime
• Flexibility

32
Role transitions
SWITCHOVER
• It is a planned and reversible change that is made to perform maintenance tasks on the system where
the primary is hosted
• The one that was primary becomes standby, and the one that was standby becomes primary
• On a switchover, no data will be lost.
FAILOVER 1/2
• A Failover is an unplanned role change. It should be performed when there is an error or unrecoverable
failure in the primary database. To continue with the service, one of the standby databases assumes the
role of new primary.
• The one that was standby becomes primary, and the one that was primary stops giving service and
being part of that Data Guard (although there are methods to recover this database as a stby again:
restore, flashback ...)
• During a Failover there may be data loss. It will depend on the existing protection settings and the error
that occurred.
• Should be used only in case of emergency

33
Role transitions
Failover 2/2
• Manual failover:
• Performed manually by the DBA
 Complete:The role change is made once all the data has been applied.
 Immediate:The role change is made without checking if data is being lost.
• Fast-start failover:
• Performed automatically by the Data Guard Broker
• When certain conditions are met (Ex: primary fall)
• Es un Complete Failover
• Exception: FastStartFailoverLagLimit. Allows data loss within the lag
•

34
Role transitions
Client connection string
• In order for client applications to connect to the new database after a role change, the
connection string must be configured in a special way.
• Always use application services. Do not use the one that brings by default the BD
•

1
2
3
4
Concepts
Types of Standbys
Protection Modes
Role transitions
35
5
Agenda
6 Configuration
7 Administration
8 Active Data Guard
9 RAC vs. DG
10 Scenarios

36
Configuration.
Creation of the DG
Cold backup
Restoration with RMAN
Creating a Physical Standby Database (Doc ID 1475344.1)
Duplicate of the DATABASE (>=10GR2)
Step by Step Guide to Create Physical Standby Using RMAN DUPLICATE (non ASM) on
different / new host (Doc ID 374069.1)
Duplicate from Active DB (>=11.2)
Creating Physical Standby Using RMAN DUPLICATE...FROM ACTIVE DATABASE [ID
1075908.1]
Cloud Control
DBCA (>=12.2)

37
Configuration.
Parameters
• Identification of BD
• DB_NAME
• DB_UNIQUE_NAME
• Sent REDO
• LOG_ARCHIVE_CONFIG='DG_CONFIG=(primary_UN,standby_UN)’
• STANDBY_FILE_MANAGEMENT=AUTO
• remote_login_passwordfile=EXCLUSIVE
• Log_archive_dest_n (El principal. Detalles a continuación)
• LOG_ARCHIVE_DEST_STATE_n
• GAPS
• FAL_SERVER=tns_primary
• FAL_CLIENT=tns_standby (11GR2>= opcional)
• Conversión paths
• DB_FILE_NAME_CONVERT
• LOG_FILE_NAME_CONVERT

38
Configuration
LOG_ARCHIVE_DEST_n 1/2
Primary
Database
chicago
Standby
Database
boston
Not used
log_archive_dest_1='LOCATION=USE_DB_RECOVERY_FILE_DEST
VALID_FOR=(ALL_LOGFILES,ALL_ROLES)
DB_UNIQUE_NAME=chicago_UN‘
*.log_archive_dest_2='SERVICE=tns_boston SYNCAFFIRM
NET_TIMEOUT=60VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE)
DB_UNIQUE_NAME=boston_UN'
log_archive_dest_1='LOCATION=USE_DB_RECOVERY_FILE_DEST
VALID_FOR=(ALL_LOGFILES,ALL_ROLES)
DB_UNIQUE_NAME=boston_UN‘
*.log_archive_dest_2='SERVICE=tns_chicago SYNCAFFIRM
NET_TIMEOUT=60VALID_FOR=(ONLINE_LOGFILES,PRIMARY_ROLE)
DB_UNIQUE_NAME=chicago_UN'

39
The attributes ofLOG_ARCHIVE_DEST_n:
• SYNC y ASYNC
- Specifies whether network I/O operations have to be performed synchronously or
asynchronously when using the LGWR
- Default ASYNC
• AFFIRM y NOAFFIRM
- Ensures that the redo was successfully written to disk (stby redo logs) on the side of the standby.
- When using ASYNC, the default is NOAFFIRM
When using SYNC the default is AFFIRM
Configuration
Transport of redo 2/2

1
2
3
4
Concepts
Types of Standbys
Protection Modes
Role transitions
40
5
Agenda
6 Configuration
7 Administration
8 Active Data Guard
9 RAC vs. DG
10 Scenarios

41
Administration
Interfaces
 Command interfaceSQL*Plus
 Data Guard Broker. DGMGRL
Tool that allows the automation and centralization of tasks such as creation, maintenance and
monitoring of Dataguard configuration. We can perform operations on the DATABASE of the machine
to which we connect, or remotely (to another DATABASE that is part of the DG) through the same
interface.
 Graphical Environment . Enterprise ManagerCloudControl (EMCC)
You have to configure the broker to access from EMCC

42
Administration
Data Guard Broker 1/5
• Automate and simplify tasks. Role changes, error monitoring, protection mode change, etc.
• TheOracle Data Guard Broker is included with EE (Enterprise Edition) and PE (Personal Edition) software
• It is also included in the client software, if you choose the "Administrator" option of the Oracle Universal
Installer

43
Administration
Prerequisites del DG Broker 2/5
Step by Step How to Create Dataguard Broker Configuration (Doc ID 984622.1)
• The Dataguard must already be configured
• both the primary database and the standby must use the same version of the database
• Spfile has to be used to allow the broker to synchronize its parameters with those of the
DATABASE
• You have to set the DB parameter to TRUE DB_BROKER_START
• You have to indicate in the db parameter DG_BROKER_CONFIG_FILEn where the dgb
configuration file is on the db server
• Register a STATIC service with the local listener of each instance (not required in RAC or
Oracle Restart)
• The primary database must be open and in archivelog mode
• The parameter of BD COMPATIBLE has to be the same in both the primary and the stby
• you must use the db parameter LOG_ARCHIVE_DEST_n with the services clause

• Disaster Protection:
• Automatiza muchas de las tareas manuales requeridas para monitorizar y configurar un Oracle
44
Administration.
Benefits of the DG Broker 3/5

45
You can use the SHOW CONFIGURATION command to display a brief description of the
configuration
Administration
DG Broker. SHOW CONFIGURATION 4/5
DGMGRL> show configuration
Configuration – DRSolution
Protection Mode: MaxPerformance
Databases:
boston - Primary database
WARNING: ORA-16809: multiple warnings detected for this
database
bostonFS - Far Sync
london - Physical standby database
london2 - Logical standby database
londonFS - Far Sync (inactive)
Fast-Start Failover: DISABLED
Configuration Status:
WARNING

46
you can use the show database verbose command to display all database properties related to the
dg:
Administration
DG Broker. SHOW DATABASE VERBOSE 5/5
DGMGRL> show database verbose london
Database - london
Role: PHYSICALSTANDBY
IntendedState: APPLY-ON
Transport Lag: 0 seconds (computed 1 second ago)
Apply Lag: 0 seconds (computed 1 second ago)
Apply Rate: 1.52 MByte/s
RealTimeQuery: OFF
Instance(s):
london
Properties:
DGConnectIdentifier = 'london'
…

• Fast-Start Failover (FSFO)):
• It allows the DGB to automatically failover to a previously chosen DATABASE in the event of loss of the primary database. No
primary database. No manual steps would be required
• This is only possible with DGB and is configured with the DGMGRL command or the EM Cloud Control
• If the DG is on Max Prot or Max Avai, this automatic failover ensures zero data loss
• in the Max Perf case, this automatic failover ensures that, in the worst case, the data indicated in seconds by the
by the FastStartFailoverLagLimit parameter of the DGB is lost at most
• If the data loss indicated by such a parameter cannot be guaranteed, then the roll generation in the primary database will be
primary database will be stopped.
• In MaxAval and Max Per, to prevent that redo freeze from lasting a long time, either the observer or the target standby can allow the primary
can allow the primary database to continue generating redo, but it will be noted that the FSFO could not be carried out.
• Observer:
• Observers are separate OCI clients that are running on servers other than primary and stby.They are responsible for monitoring
responsible for monitoring the BD primary. (In 19c you can have up to 3 in Active/Passive mode for HA)
• If both the "observer" and the target stby lose connectivity to the primary DATABASE (for a period of time indicated in the
indicated in the FastStartFailoverThreshold DGB property), the observer tells the DGB to fire an FSFO to the BD stby.
47
Administración
DG Broker. Fast-Start Failover 1/2

48
Administration
DG Broker. Fast-Start Failover andObserver 2/2
The DG Broker is able to
automatically reinstate the previous
primary as soon as that "dropped BD"
is mounted and communications have
been restored throughout the DG
(including observer)

49
With Cloud Control we can:
Create a standby
Make a Role Change
Monitor dg status
Change the protection mode
Add standbys
Administration
CLOUD CONTROL

1
2
3
4
Concepts
Types of Standbys
Protection Modes
Role transitions
50
5
Agenda
6 Configuration
7 Administration
8 Active Data Guard
9 RAC vs. DG
10 Scenarios

51
Active Data Guard (ADG)
The Active Dataguard (ADG) option is available starting with version 11.1 and requires
additional licensing.
The physical standby is in a read-only state and applying changes. This allows us to
have a synchronized environment where you can run ONLINE reporting while still
protected against disasters.
This functionality is also called Real-Time Query
The COMPATIBLE must be set at least to 11.0.0 in both the primary and standby
http://www.oracle.com/technetwork/database/availability/active-data-guard-wp-12c-
1896127.pdf

Standby in
Dataguard
Standby inActive
Dataguard
Open mode MOUNT READ-ONLY
Reporting OFFLINE ONLINE
Features • Real time apply • RealTime apply
52
Active Data Guard
DG vs ADG

53
• Oracle processes can validate data from the redo BEFORE it is applied to the stby
•
• LOST-WRITE DETECTION
- A lost-write occurs when the I/O system confirms that a write has been made, but
has not actually been performed on disk.
- "lost-write detection" can be implemented with the parameter
DB_LOST_WRITE_PROTECT=TYPICAL in the primary and standby
-
• AUTOMATIC BLOCK REPAIR
- When a data corruption is detected, it is able to re-order valid redo logs to get the
corrected block.
-
• Best Practices for Corruption Detection, Prevention, and Automatic Repair - in a Data Guard Configuration
(Doc ID 1302539.1)
Active Data Guard
Prevention of data corruption 1/2

54
Active Data Guard
Automatic Block Repair 2/2
Primary Standby 2
Block
• From 11gR2, the BD primary automatically tries to repair the corrupt block (which it has
detected) in real time by bringing a good version of the same block from the physical stby. In
the same way it happens in the stby in the case of an ADG, but this time bringing it from the
primary.
• Automatic block repair reduces data downtime due to block corruptions.
• Automatic block repair requires ADG to be used
•

55
Active Data Guard
DML in Global Temporary Tables 1/2
From 12c we can launch some DML operations in the standby: Insertions can be made in
global temporary tables.
In order to use this functionality we must have Active Data Guard.
From 12c the parameter TEMP_UNDO_ENABLED appears for the BD Primary, which has
False as the default value. Its activation allows temporary tables to use a specific UNDO,
improving performance and reducing the size of the generated REDO.
In this way the redo generated by temporary tables does not have to be transmitted to
the stby.

56
Active Data Guard
DMLs from standby 2/2
From 19c we can launch DML operations in the standby (inserts, update, deletes), not
only in global temporary tables.
In order to use this functionality we must have Active Data Guard.
DB parameter ADG_REDIRECT_DML=TRUE
The DML statement is launched from the standby, executed on the primary, and the data is
returned to the stby.
Until the commit is made in the primary, the result of this opeación is only visible by the stby that
executed it, not by the others.
As soon as the commit is made (it is actually done in the primary) it can already be seen by all
the stbys.
It's not very scalable by the fact that it actually occurs in the primary, so it's best to avoid massive
DMLs. This is an option to be used punctually.

57
Active Data Guard
Sequences from standby
The Global sequences created in the Primary can be consulted from the stby
The sequence numbers assigned will be UNIQUE throughout the DG
Session Sequence:
For Global Temporary Tables
Only visible by that session
The sequence number will be unique for that session, but NOT between sessions
Use:
For reporting offloading that require unique sequence numbers

1
2
3
4
Concepts
Types of Standbys
Protection Modes
Role transitions
58
5
Agenda
6 Configuration
7 Administration
8 Active Data Guard
9 RAC vs. DG
10 Scenarios

59
DG vs RAC 1/3
VS
RAC Data Guard
RAC Extendido RAC Local + DG
MAA with Oracle 19c:
https://www.oracle.com/a/tech/docs/maximum-
availability-wp-19c.pdf

60
DG vs RAC 2/3
: better : Poor : N/A RAC local +DG RAC widespread
Availability
Compute scalability
Taking advantage of environments Crusader
HW and SW maintenance and patching Standby-first and Rolling Rolling
Migrations and upgrades Standby-first and Rolling Rolling
Data corruption and unrecoverable database drops
Backup window From the standby Assign a node
Problems of communications between CPDs
asynchronous
synchronous
Drop in communications between DPCs
asynchronous
synchronous
Fall of a CPD
Exadata support

61
DG vs. RAC
RTO/RPO 3/3
RAC local + DG RAC widespread
Downtime-
RTO
Potential Data
Loss – RPO
Downtime-
RTO
Potential Data
Loss - RPO
Disk failiure Zero Zero Zero Zero
Machine or DB instance crash Zero to seconds Zero Zero to seconds Zero
Data corruption or unrecoverable BD losses
Seconds to
minutes
Zero to seconds Hours Since last backup
CPD Drop
Seconds to
minutes
Zero to seconds Zero Zero
Database reorganization, file movement, one-off patching Zero Zero Zero Zero
Maintenance of hardware, software. Patching Zero to minutes Zero Zero to hours Zero
DB Upgrade (Both patch sets and full releases)
Seconds to
minutes
Zero
Minutes to
hours
Zero
Upgrades of the objects of the application (tables...). Hours Zero Hours Zero

62
Let's recap
Data Guard = Data Replication++

63
Let's recap
Tipos de Standby:
• Physical
• Snapshot
• Active Data Guard
• Cascaded
• Far Sync
• Logical
Protection Modes:
• Max Yield
• ASYNC NOAFFIRM
• Max Protección
• SYNC AFFIRM
• Max Availability
• SYNC AFFIRM
• SYNC NOAFFIRM

64
Let's recap
Switchover Failover
Role transitions

65
Let's recap
Data Guard Broker Observer & Fast-Start Failover

66
Let's recap
Active Data Guard (licenciamiento adicional)

67

Descanso
68 Copyright © 2022 and/or its affiliates. All rights reserved.Confidential | Oracle Public

1
2
3
4
Concepts
Types of Standbys
Protection Modes
Role transitions
69
5
Agenda
6 Configuration
7 Administration
8 Active Data Guard
9 RAC vs. DG
10 Scenarios

Scenarios
Reporting andValidations
Backup
Tuning
Plans of Disaster Recovery
Upgrades with minimal downtime
Migrations with minimal downtime
Data Guard and Multitenant
Copyright © 2022 and/or its affiliates. All rights reserved.Confidenti
70

Decrease primary workload
Copyright © 2022 and/or its affiliates. All rights reserved.Confidential | Oracle
Public 71
Scenarios.
Reporting and Validations
71

72
Scenarios
Reporting and Validations
Reporting (read-only)
• Reporting Offline. Dataguard normal
• Reporting Online. Active Dataguard
• Reporting Online. Standby Lógica
We can access the standby database to launch heavy queries, classify, generate reports,
access via the web, etc. and free up our production environment thus improving
performance.
Validations or load tests (read-write)
Snapshot standby
Perform validations of our contingency plan.
Perform load tests without penalizing the performance of our production service.

73
Reporting (No up-to-date data )
ADG is not used. The standby only stays synchronized when it is mounted, so
mounted, so we will have to open it in read-only to be able to consult it and
consult it and during that time the data will not be updated.
Scenarios
Reporting. ReportingOffline

74
ReportingOnline.Active Dataguard. (Updated Data )
By default, an open standby is accessed in read-only mode with the updated data until the moment of the
the moment of the start of the process, from version 11G, using Active Dataguard*, synchronization can be
synchronization can be maintained throughout the process and the data accessed will be identical to those
be identical to those of production.
* Active Dataguard, es una funcionalidad que necesitalicenciamiento.
Scenarios
Reporting. ReportingOnline.Active Dataguard

75
ReportingOnline a a logical standby
A logical standby is created from a physics.
Open in writing with restrictions
It has to meet requirements to be Logical, less flexibility.
flexibility.
Scenarios
Reporting. ReportingOnline. LogicalStandby

76
Scenarios
Validations and Load Tests
With this procedure we will have our standby database in read-write mode. As the
snapshot standby is still a type of standby, it will continue to receive the files correctly
although it will not apply them until it returns to physical standby.
Prerequisites:
 You have to configure the Flash Recovery Area
 The standby does NOT have to have flashback configured, although it will use that
functionality during the procedure.
 The files will be received from primary school but will not be applied until it is again
physical standby

Decrease primary workload
Copyright © 2022 and/or its affiliates. All rights
reserved.Confidential | Oracle Public 77
Scenarios.
Backup
77

78
Scenarios.
Backup. Backup from the standby 1/2
How to take consistent backups at standby site (Doc ID 1292126.1)
• The backups that are made in the physical standby can be used to recover the primary or vice versa.
• Launching the backup in the standby will give us flexibility when selecting our backup plan.
– Example. We can make FULL daily even if these take a long time and get into the production window.
• Only with Physical Standby. This is not valid for Logical Standby
• An RMAN catalog is required. Backup differs by DB_UNIQUE_NAME
•
• Spfile is not interchangeable
• You can configure block change
tracking in the standby also to
accelerate the backups (Requires
ADG license)

79
Scenarios
Backup from standby. Archivelog retention 2/2
If we want to continue launching them in the primary, with a correct configuration we will
not need to change our backup policy.
In the primary:
We configure the retention of the archives:
CONFIGURE ARCHIVELOG DELETION POLICY TO SHIPPED TO ALL STANDBY;
CONFIGURE ARCHIVELOG DELETION POLICY TO APPLIED ON ALL STANDBY;
In the stby:
CONFIGURE ARCHIVELOG DELETION POLICY BACKED UP n TIMES TO DEVICE TYPE SBT;
RMAN "backup database archivelog delete input" will not delete the archives that
are still required for synchronization.

Decrease primary workload
reserved.Confidential | Oracle Public
80
Scenarios.
Tuning
80

81
Scenarios
Tuning. Diagnostics Pack & Tuning Advisor
• ADG license required and Tuning Pack
• Primary Performance Analysis
• The Oracle Diagnostics Pack can be used with ADG, which allows you to evaluate AWR & ADDM
ADDM
• The stby is required to be open in Read-Only
• We can also run the SQL Tuning advisor (needs license)
• Keep in mind that the AWR in the standby is an "exact copy" of the primary

Security and Availability

82
Scenarios.
Disaster Recovery Plans (DR))
82

83
Failover
• In case of disaster, total loss of the production database
• Loss of data in the event that all the redo has not been transferred to the stby and that
Scenarios.
DR Plans. Production Server Loss
PRIMARIA STANDBY

Protection against corruption
84
Scenarios
DataSecurity
84

85
Physical corruptions. Do not replicate
Ability to recover datafiles without using backup:
Recovering the primary database's datafile using the physical standby, and vice
standby, and vice versa (Doc ID 453153.1)
LogicalCorruptions. Detected
Possibility of detecting "Lost-write" type corruptions that occurred in our
occurred in our production database thanks to the existence of a physical
a physical standby.
Alert standby...
STANDBY REDO APPLICATION HAS DETECTED THAT THE
PRIMARY DATABASE LOST A DISK WRITE OF BLOCK 26, FILE 7
https://docs.oracle.com/en/database/oracle/oracle-database/19/sbydb/examples-of-using-oracle-data-
using-oracle-data-guard.html#GUID-8F4E7807-6013-480F-8780-088F5639732F
Scenarios
Data security. Corruptions

86
Perrequisite -> The Physical Standby must have active flashback database functionality,
the time of deletion must be within the defined
db_flashback_retention_target and the archive/flashback logsAvailable
Scenarios
Data security. Recovering the deletion of an object
PRIMARIA STANDBY
SQL > flashback database to timestamp
to_date(‘12-MAY-2014 14:30:00','DD-MON-
YYYY HH24:MI:SS');
Read only
flashback on
Drop/Truncate/D
elete
EXPDP

Minimize downtime
Minimize BUG regressions

87
Scenarios
Data Guard Maintenance
87

88
Scenarios
DG maintenance. Planned shutdown of the production server
Switchover (role change)
• In case of need for hardware/software upgrade
• No Data Loss.
• Reversible change.There must be visibility between the two
•

89
Scenarios.
DG Maintenance. Patching Options
•
• Conventional procedure
•
With stop
time
• >=11.1 Transient logical standby
• >=11.2 Oracle Data Guard Standby-First Patch
changes)
• >=12.1 DBMS_ROLLING
Mminimum
downtime
(switchovers)
https://www.oracle.com/technetwork/database/availability/database-rolling-upgrade-3206539.pdf

90
Scenarios
Maintenance of the DG. Planned shutdown. Conventional upgrade
Shutdown services
./runInstaller (installation new version)
SQL>startup mount;
SQL>alter database recover managed
standby database disconnect using
current logfile;
Tiempo
parada
+/- 60 ‘
Shutdown services
./runInstaller (installation new version)
SQL> startup
SQL> @?/rdbms/admin/catpatch.sql
1
2

• Transient Logical Standby. A Logical Standby is used temporarily (the Physical Stby becomes
Logical stby)
• The fact that a Transient Logical Standby is used is because, being open in writing, it allows you to
upgrade it. You can modify the catalogs of the DATABASE without touching the objects of the
application.
• >= 11GR1
• Proceeds:
• Improve availability by minimizing scheduled downtime. Very useful in production systems
without HA architecture (RAC)
• The catalog upgrade only has to run on 1 database
• Additional validations can be performed before switchover
• When the process finishes both databases are in the same version
91
Scenarios
Maintenance of the DG. Minimum stop. RollingUpgrade 1/3
91

92
Scenarios
Maintenance of the DG. Minimum stop. Rolling Upgrade 2/3
• Prerequisites:
• A Dataguard environment with a physical standby
• Requirements for logical standby must be met
• If Broker is being used, it is recommended to disable it during the procedure.
• The protection mode must be Maximum Availability or Maximum Performance.

93
Scenarios
Maintenance DG. Minimum stop. Rolling Upgrade 3/3
1
2
3
Automatic
procedure
physru (script)
Using Transient Logical Rolling Upgrade for Database
Migration (Doc ID 2350945.1)

94
Scenarios
• Package DBMS_ROLLING:
– >=12.1
– It allows us to upgrade in
Rolling mode, minimizing
downtime (similar to physru but
more automated).
– Also uses Logical Stbys
temporarily
– Especially recommended for
complex scenarios with multiple
stbys
– Rolling upgrade using DBMS_ROLLING -
Complete Reference (Doc ID 2086512.1)

95
Scenarios
• 2 groups of databases to be considered in
the DG:
• Leading group (LG)
• These databases are the first to upgrade
(before swithover)
• The LG has a master DB (the future primary)
• Trailing group (TG)
• These databases are the ones that will be
updated the last ones (after the switchover)
• The TG has a master DB (the original primary)
• El sentido de hacer grupos es para que ninguna de
las BBDD se quede desprotegida
• Este procedimiento es capaz de manejar fallos en
el proceso de upgrade de alguna de ellas,
pudiendo hacer failovers si fuese necesario.

96
Scenarios
• DBMS_ROLLING steps:
• exec dbms_rolling.init_plan(future_primary => orcl12dg’)
• exec dbms_rolling.build_plan();
• exec dbms_rolling.start_plan();
• UPGRADE MANUAL/DBUA standby
• exec dbms_rolling.switchover;
• UPGRADE MANUAL/DBUA nueva standby;
• exec dbms_rolling.finish_plan;

97
Scenarios
Example of a plan generated by the DBMS_ROLLING:

98
Scenarios
Maintenance of the DG. Minimum stop. Standby First Apply 1/2
• Para >= 11gR2
• RUs, RURs y
one-offs

99
Scenarios
Maintenance of the DG. Minimum stop. Standby First Apply 2/2
PROCEDURE
Step 1 We apply the patch on the standby. ( Mixed environment DG physical supported
temporarily. 31 days)
Step 2  We validate the patch.
We can convert the physical standby into snapshot standby (read-write) perform the
post-installation steps of the patch and validate the operation after the changes.
Step 3  Normalize the environment
Opcion 1: We apply the patch in the primary (It implies stop if it is not RAC)
Opcion 2: We perform a switchover and apply the patch in the new standby, and
execute the SQL changes in the primary. We will do another switchover if
we wanted to get back to the situation actual.
Opcion 3: We rollback the patch at the standby

 Minimize downtime...
* 3 min stop + reset connections
100
Scenarios.
Migrations taking advantage of the DG
100

101
Minimum downtime:
Server Migration
Architecture Migration
Storage Migration
Platform Migration
Version Migration
Scenarios
Migrations taking advantage of the DG

102
Scenarios.
Migrations leveraging DG. Server Migration 1/3
Source and destination server must share the following characteristics:
Same Platform
SameVersion and Patch Set
Supported differences:
Hardware Manufacturer
HardwareConfiguration
Processor
OperatingSystem Distributor
Operating system version

Standby Physics!!

103
Scenarios
Validation Phase (Optional))
As a previous step to the switchover, we could even perform a battery of tests in the new standby before taking
standby before taking the final step and migrating to the new machine.
Standby snapshot

104
Scenarios
Final Phase (Switchover or Failover)
Move the primary to the new machine and get rid of the standby on the old machine

105
Scenarios
Migrations taking advantage of the DG. Db architecture change
From RAC to Single Instance
MAA - Creating a Single Instance Physical
Standby for a RAC Primary [ID 387339.1]
From Single Instance to RAC
Migrating to RAC using Data Guard [ID
273015.1]

106
Scenarios
Migrations taking advantage of DG. Storage change
From FILE SYSTEM to ASM
Creating ASM physical standby from non
ASM primary [ID 752360.1]
From ASM to FILE SYSTEM
Step-By-Step Guide To Create Physical
Standby On Normal File System For ASM
Primary using RMAN [ID 838828.1]
DB_FILE_NAME_CONVERT
LOG_FILE_NAME_CONVER
T
LOG_ARCHIVE_DEST_n

107
Scenarios
Migrations taking advantage of dg. Platform change
Migration Of An Oracle Database Across OS Platforms
(Generic Platform) (Doc ID 733205.1)
• Export / Import to include the use of Datapump facilities.
facilities. All versions support Export/Import, but for
Datapump, 10.1.0.2 or higher is required
• Transportable Tablespaces 10G or Later
• RMAN Convert Database functions. 10G or Later
• Streams Replication
• Create Table As Select (CTAS)
• Dataguard Heterogeneous Primary and Physical Standbys*
• Oracle Golden Gate
Data Guard Support for Heterogeneous Primary and Physical Standbys in Same Data Guard Configuration (Doc ID 413484.1)

108
Scenarios
Migration taking advantage of DG. BD 1/6 version change
Pre-requisitions rest scenarios
Server Migration
Architecture Migration?*
Platform Migration? *
*Data Guard Support for Heterogeneous Primary and Logical Standbys in Same Data Guard Configuration
[ID 1085687.1]
Mixed Oracle Version support with Data Guard Redo Transport Services (Doc ID 785347.1)

109
Scenarios
Migration taking advantage of DG. Db 2/6 version change
• Example: You have a database in version 12c and you want to migrate to 19c
• Take advantage of the fact that a Logical Standby is open in RW to upgrade to it and that
a smaller version can send redo to a version BD major
– On the contrary, it is not allowed (sending redo from a major version to a minor one),
therefore a role change could be made, but the new standby could not be opened. It
would be necessary to make the UPGRADE before opening it
– All the prerequisites necessary to mount a Logical Standby must be met
• Minimum downtime on migration
•

110
Scenarios
Migration taking advantage of the DG. Change of version of BD.
Steps. 3/6
Version migration requirements:
 We start from a single database without DG that must
be migrated
 A physical Stby is created and transformed to Logical
Stby
• All records must have unique identification
• Verify data types and objects
 Enable Flashback Database
 Disable the Broker during the upgrade process
 Review the transport configuration so that there are
no switchover errors
 Validate the proper functioning of the application in
the new version by previously migrating a test
environment


111
Scenarios
Steps 4/6
 We stop the transport of redo
 We upgrade at the standby
• Dbua
• Manual Upgrade
• …

112
Scenarios
Steps. 5/6
 Mixed Environment.
Example
Primary 12c
Logical Standby 19c
 We activate the redo transport and
synchronize
 We validate the new environment

113
Scenarios
Steps 6/6
We made a switchover and started to
service with the new database in 19c (small
service outage)

 2 Options from here:
 We raise the old primary with the
new 19c binaries and activate the SQL
apply to synchronize
We deleted the old primary because
we no longer need it


115
Scenarios
Data Guard con Multitenant
• A CBD can contain PDBs that are Physical Stbys, Logical Stbys, and Snapshot Stbys
• The role of the DATABASE is defined only at the CBD level, NOT the PDB level
• Individual PDBs have no role of their own.
• Role transitions run at the CBD, NOT PDB level
• It is not necessary to have the same number of PDBs in the primary CBD as in the standby
• if a new pdb is created in the primary cbd it is automatically replicated to the stby (this behavior can
be modified by command)
Multitenant (more than 3 user
PDBs on a CBD) requires
additional license

Data Guard 19c Data Guard 19c Data Guard 19c

More Related Content

What's hot (20)

Similar to Data Guard 19c Data Guard 19c Data Guard 19c (20)

Recently uploaded (20)

Data Guard 19c Data Guard 19c Data Guard 19c