Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Oracle Ac?ve Data Guard: Power, Speed, Ease, and Protec?on CON6531 Larry M. Carpenter Master Product Manager Oracle High Availability Systems
Suresh Torlikonda Weblogic SOA Infra Architect
Vasu Raghunathan Director, Database Administra?on
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
• Available now – Exadata Express Cloud Service
• Coming soon – Database Cloud Services – Exadata Cloud Machine
2
Announcing Oracle Database 12c Release 2 on Oracle Cloud
Oracle is presen?ng features for Oracle Database 12c Release 2 on Oracle Cloud. We will announce availability of the On-‐Prem release some?me aVer Open World.
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Safe Harbor Statement The following is intended to outline our general product direc?on. It is intended for informa?on purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or func?onality, and should not be relied upon in making purchasing decisions. The development, release, and ?ming of any features or func?onality described for Oracle’s products remains at the sole discre?on of Oracle.
3
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Agenda
Introduc?ons
MAA in Produc?on at Mul?Plan
Ease
Power
Speed
Protec?on
MAA in Produc?on at IndianOil
4
In all things Oracle Data Guard!
MultiPlan proprietary ─ do not distribute without permission MultiPlan proprietary ─ do not distribute without permission
Maximum Availability Architecture Suresh Torlikonda/Vasu Raghunathan September 2016
*
Goal
Regardless of the type of failure:
No noticeable downtime
All applications fail gracefully
Zero data loss
! ! !
*
• Lack of high availability for applications, databases and business processes
• Lack of seamless disaster recovery due to various site specific configurations and deployments
• Longer RTO to switchover between active and standby database
• Lack of Redundancy across all layers was causing disruption for business processes
• Single point of failures for certain applications
Problem
*
• Moved away from tradition DNS round robin to connect to RAC cluster for database
• Configure F5 LTM to load balance requests between the Web Tier to achieve another layer of
abstraction between application and web tier
• Configured F5 LTM with scan IP’s for both the active and standby database
• Configured F5 LTM to periodically check database status
• Deployed a dedicated service on the database for F5 LTM to execute a monitor SQL
• LTM executes the SQL monitor script every 30 seconds
• F5 LTM would failover to the standby database after three consecutive failures occur when
executing SQL monitor script
• By using F5 as abstract layer helped us avoid changing client configuration between switchovers and
failover
• Implementation of GTM to avoid site specific configurations
• Deployed GTM on both the sites.
• Configured GTM to resolve to the client request based on the location where the requests are
originated .
Solution/Innovation
• Introduce redundancy at all layers – Network, Storage, Server(Physical & VM), Database &
Application
• Update the configuration at the existing Disaster Recovery Datacenter to match production
which is located approximately 1500 miles apart
• Utilize technologies and practices to maintain infrastructure, deployment and code level
across datacenters
• Introduce service names instead of site specific hostnames in configuration to establish
communication between various components in data centers
• Identical URI’s in both locations to access the services
Solution Innovation
LTM Local Transaction Manager is used to proxy and load balance request across multiple hosts with a single address
URI Uniform Resource Identifier is a mechanism of retrieving a resource existing on intranet or extranet network
GTM Global Transaction Manager is used to resolved names to IP’s based on availability and location
* 10
The diagram below shows how our conceptual architecture is mapped onto the Fusion Middleware stack
Ø Service Bus, SOA Suite, Weblogic ,Coherence, JMS
SOA Suite Composite Services
Service Bus
Business Services
BPM, BPEL, Components
Business Rules Components
Weblogic Services
Service Implementations
Messaging
Coherence
Business Services
External Requests
Oracle
Proxy Services Proxy Services Proxy Services
MDS Repo SQL Server
Ø Virtualization
Ø Orchestration Ø Implementation
Architecture
*
• Database
• OS – Oracle Enterprise Linux – 6.5
• ASM/Grid – Oracle 12.1.0.2
• Database – Oracle 11.2.0.4, 12.1.0.2
• SQL Server 2008, 2012, 2014
• Middleware
• Oracle Service Bus -12.1.3
• Weblogic Application Server -10.3.6/12.2.0
• Coherence - 3.7/12.2.1
• WebSphere - 6.x
• Cold Fusion - 10.x
Platform Components
*
• Maximum availability for local failover/switchover – with zero Data loss
• Local Standby - SYNC Mode
• Used for patching, this reduces the production downtime
• Read only queries
• DR Standby – ASYNC Mode
• Moved RMAN backups to DR – Freeing up I/O and CPU resources on primary
• We generate close to 1TB archive logs per day, the lag between Primary and DR standby is less than a few minutes
Active Data Guard Setup
*
Netw
orkWe
b Tie
r
Web
Tier
App
Tier
App
Tier
Datab
ase
Tier
Datab
ase
Tier
Prod
Stand
by
Prod
LOCAL DNSLOCAL DNS
1) The client machine asks the local DNS server the question: what IP is www.abc.com?
2) The local DNS server realizes www.abc.com is a CNAME
3) The DNS server recurses to the GTM for the question: what is the destination for www.abc.com?
Site 1
GTM 4) The GTM reviews the topology records and determine either Site1 or Site 2
is the active datacenter for that www.abc.com and returns the virtual server LTM VIP from Site 1 or Site 2
Netw
ork
Storage Array 1 Storage Array 2 Storage Array 3
Active Data Guard
Site 2
Multi Site
*
Service to Database
Application Servers
Request Flow
Local DNS GTM
LTM
Service to Service
Application/Web Servers GTM Local DNS
Client
Local DNS GTM
LTM
Client to Service
*
• Using F5 as abstract layer reduced the RTO from 2hrs to less than 5 min for failover and less than 15 min for manual
switchover within a site
• The actual time for failover will be reduced once the Observer is turned on
• Combination of GTM and LTM reduced the time to cutover to a Disaster Recovery site to less than hour
• By utilizing network devices like (GTM and LTM) the configurations were maintained at single locations for multiple
datacenters
• Automated release management to eliminate the need of deployments to multiple sites
Benefits
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Thank you! Back to you Larry
17
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
EASE Making your life easier
18
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Ease of use in Data Guard Release 12.2 • Use DBCA and EMCLI to create Standbys • Use RMAN and Enterprise Manager to create Far Sync Instances • Rest Interface to the Data Guard Broker • Chef tools to create and manage Data Guard standbys • Puang the FSFO Observer in the Background • Scrip?ng with the Broker DGMGRL command line interface
19
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
DBCA Standby Crea?on
• At the Standby server install the Oracle SoVware and start a listener • Execute a single line command
• First release caveats – SIDB only, does not support Mul?tenant Primary databases, yet! – Cannot directly create RAC Standby databases, but they can be converted to RAC.
20
Simple and scriptable!
dbca -silent -createDuplicateDB -primaryDBConnectionString myprimary.domain:1523/chicago.domain -gdbName chicago.domain -sid boston -initParams instance_name=boston -createAsStandby -customScripts /tmp/test.sql
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
DBCA Standby Crea?on
• Use the post SQL to ?dy things up (in /tmp/Test.sql)
• Then configure your network TNSNAMES for Data Guard • And Build your Broker configura?on.
– Or just add this new standby to your exis?ng Broker configura?on
21
Finish up
connect / as sysdba alter database flashback on; shutdown immediate startup alter system register; exit
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Enterprise Manager Cloud Control
• Enterprise Manager has had a Standby Crea?on Wizard for many versions • It also has a command line interface called EMCLI
– Now you can script standby crea?on through it.
– No?ce how the target can be a RAC or Single Instance database – RMAN duplicate from ac?ve database is also the default
22
Command line standby creaCon
emcli create_standby -source_db_target_name=“BOS” -source_db_target_type=“rac_database” -dest_oracle_sid=“BOS1_1” -spname=“BOS1” -use_broker çMy favorite qualifier! (it’s the default too!)
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. | 23
URL access to DGMGRL funcCons and more! A RESTful API for Oracle Data Guard Broker
Primary Data Center DR Data Center
Data Guard Redo Transport
Broker Service
Broker Service
HTTP(S) RESTful Services Interfaces
DBCA Service
DBCA Service
Data Guard Broker
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
A RESTful API for Oracle Data Guard Broker
24
Convert back to Physical Standby from Snapshot Standby
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Chef1 for Data Guard
• Automate your Physical Standby Crea?on – Chef as solu?on to automate OS & database configura?on – Chef cookbook/recipes designed for flexible deployments – Cookbook akributes designed for Data Guard setup between:
• DBaaS <-‐> DBaaS • RACDBaaS <-‐> RACDBaaS • ExaCS <-‐> ExaCS • On-‐prem (standalone) -‐> DBaaS • On-‐prem (RAC) -‐> RACDBaaS • On-‐prem (RAC) -‐> ExaCS
25
Now We’re Cooking
1hkps://www.chef.io/chef/
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Automa?ng Data Guard
• Chef cookbook process op?ons: – Data Guard setup using duplica?on from Ac?ve Database – Data Guard setup using duplica?on from local backup – All environments configura?ons and process op?ons available through cookbook akributes configura?on
26
Chef cookbook
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Automa?ng Data Guard
• Download the cookbook1 tar (soon) and upload to your chef server • Configure cookbook akributes to match system specific configura?ons
– For flexibility purpose there is an op?on to provide a json file with most frequent changes in configura?on
• Execute a single recipe on the primary site • Execute a single recipe on the standby site • Execute a single recipe to enable Ac?ve Data Guard.
27
Deployment
1Stay Tuned to Oracle.com for more informa?on
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Power Giving you more control and ability
28
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
More Power in Data Guard in Data Guard Release 12.2
• Using all your Standby Instance to apply redo • Easily create Subset Standby databases in Mul?tenant • Data Guard Broker Support and involvement in DBMS_ROLLING controlled rolling upgrades
• Migrate and Failover PDBs with the Broker
29
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Physical Standby Redo Apply • The MRP and its redo apply servers have always run on one node of a Physical Standby RAC.
• This limits it to the CPU and IO power of that one node. • But s?ll our performance has been geang beker with each release.
30
Standby Apply Rate
MB/sec
0
200
400
600
800
Oracle 9i Oracle 10g Oracle 11g (Commodity)
Oracle 11g (Exadata)
OLTP Workload
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Single-‐Instance Redo Apply
MRP Processes
MRP Processes
Standby Instance 1
RFS Process
Coordinator Process
MRP Processes
RFS Process
Coordinator Process
MRP Processes
RFS Process
Primary Instance 1
ASYNC/SYNC Process
Primary Instance 2
ASYNC/SYNC Process
Primary Instance 3
ASYNC/SYNC Process
Thread 1 Redo
Thread 2 Redo
Thread 3 Redo
SRL
SRL
SRL
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Mul?-‐Instance Redo Apply – 3 Node Standby
Coordinator Process
MRP Processes
Standby Instance 1
RFS Process
Coordinator Process
MRP Processes
Standby Instance 2
RFS Process
Coordinator Process
MRP Processes
Standby Instance 3
RFS Process
Primary Instance 1
ASYNC/SYNC Process
Primary Instance 2
ASYNC/SYNC Process
Primary Instance 3
ASYNC/SYNC Process
Thread 1 Redo
Thread 2 Redo
Thread 3 Redo
SRL
SRL
SRL
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Mul?-‐Instance Redo Apply – 2 Node Standby
Coordinator Process
MRP Processes
Standby Instance 1
RFS Process
Coordinator Process
MRP Processes
Standby Instance 2
RFS Process
Coordinator Process
RFS Process
Primary Instance 1
ASYNC/SYNC Process
Primary Instance 2
ASYNC/SYNC Process
Primary Instance 3
ASYNC/SYNC Process
Thread 1 Redo
Thread 2 Redo
Thread 3 Redo
SRL
SRL
SRL
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Mul?-‐Instance Redo Apply
• Parallel, mul?-‐instance recovery : standby will keep up – Standby recovery -‐ u?lizes CPU and IO across all nodes of RAC standby – Some of our OLTP workload tests on Exadata show great scalability
34
0 200 400 600 800 1000 1200 1400
1 Instance 2 Instances 4 Instances 8 Instances
OLTP Workload Standby Apply Rate
MB/sec
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Mul?-‐Instance Redo Apply
• Mul?-‐Instance Apply runs on all MOUNTED instances or all OPEN Instances
• Exposed in the Broker with the ‘ApplyInstances’ property on a standby database
• First Release caveats – Disallows the use of In Memory Column Store on the Standby – RMAN Block Change Tracking file disabled
35
recover managed standby database disconnect using instances 4;
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Oracle Data Guard and Mul?tenant • Redo genera?on is at the Container (CDB) level
– Not the Pluggable (PDB) level • Hence Data Guard works at the CDB level also. • Switchover and Failover will move the en?re CDB • Mul?ple manual steps can be taken to move a single PDB using the normal UNPLUG and CREATE PLUGGABLE commands in SQLPlus.
• Enter the Data Guard Broker! – New command to Migrate or Failover a PDB
• Also coming in Oracle Database 12.1.0.2
36
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Migrate or Failover – Same Command • Migrate a PDB from the Primary CDB to another Primary CDB or Failover a PDB from a Standby CDB to a Primary CDB
• The role of the CDB you connect to determines what happens – Connect and Execute the command on a Primary
• You migrate the REGION1 to another Primary CDB
– Connect and Execute the command on a Standby • You failover the REGION1 to another Primary CDB
37
DGMGRL> MIGRATE PLUGGABLE DATABASE REGION1 TO CONTAINER NORTH_SALES_NEW USING REGION1.xml CONNECT AS sys/mypassword@NORTH_SALES_NEW;
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Server1
CDB1 CDB2’
Server2
CDB1’ CDB2
Data Guard Broker Enhancements for Mul?tenant
• Two servers with two CDBs in each
• DG replica?on between CDB pairs in opposite direc?ons
• PDB Failure in primary
• Move corresponding PDB from standby to primary CDB in same server – Effec?vely perform failover
SupporCng individual PDB-‐level failover
PDB1 PDB2 PDB3 PDB4 PDB5
PDB1 PDB2 PDB3 PDB4 PDB5 Data Guard Rep
lica?
on
Data Guard Rep
lica?
on
Primary CDB 1
Primary CDB 2 Standby CDB 1
Standby CDB 2
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Opera?onal Requirements • The Target CDB must be on the same system
– No Remote PDB movement is possible with the Data Guard Broker
• The source PDB must be on shared storage with the Target CDB. – No data file copying is done by the Broker.
• The plugin at the target CDB is done with STANDBYS=NONE – Any standbys of the Target CDB must be manually updated with the new data files
• The source PDB will be removed from the source CDB. • Remote PDB reloca?on can be done in 12.2 with the new RELOCATE command outside the Data Guard Broker.
39
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Speed Making things go faster and more transparent
40
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Speed up Performance and Role Changes in Data Guard Release 12.2
• Use the In Memory Column Store on your Standby • Use AWR and SQL Tuning Advisor to diagnose and tune Redo Apply and Query issues on your Ac?ve Data Guard Standby
• Move users transparently during role change with Session Draining on the Primary and Preserving exis?ng connec?ons at the Ac?ve Data Guard Standby
41
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Database In-‐Memory and Ac?ve Data Guard • In-‐memory DB on Ac?ve Data Guard
– Create IMC tables and columns for analy?cs on Ac?ve Data Guard – Can populate with different data than produc?on database
• Offload even more to your standby! • First Release Caveats
Cannot be used with a standby running Mul?-‐Instance Redo Apply, yet!
42
Month In-‐Memory
Year In-‐Memory
Standby Primary
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Configuring In Memory and Ac?ve Data Guard
• Set the INMEMORY_SIZE ini?aliza?on parameter – Set on Primary Only – No Standby In Memory Column Store – Set on Standby Only – No Primary In Memory Column Store – Set on Primary and Standby – In Memory Column Store on both! (Or more)
• Set the INMEMORY akribute on all objects you want In Memory • Depending on the INMEMORY_SIZE parameter seang, the same objects will be loaded into memory at the Primary, the Standby, or both.
43
To Be In Memory or Not To Be In Memory
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Configuring In Memory and Ac?ve Data Guard
• You can control where objects are populated – On the Primary, the Standby, or both
• And have different data in memory at the Primary and Standby
• Configure role based services on the Primary and the Standby for the column store objects and start them
• Set the INMEMORY ... DISTRIBUTE FOR SERVICE clause on all objects to be populated using those services – The In Memory Column Store is populated for that object where the service runs
• Primary only, Standby only, or both.
44
CreaCng Different In Memory Column Stores
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Things to think about
• Opera?onal Restric?ons – In-‐Memory Expressions are captured based on queries executed on the primary only – In-‐Memory Informa?on Lifecycle Management (ILM) polices based on access criteria are triggered based only on access recorded on the primary database.
• Usage Restric?ons – In-‐Memory FastStart and In-‐Memory Join-‐Groups are not supported in an Ac?ve Data Guard environment.
– Using the DISTRIBUTE FOR SERVICE akribute does not allow you to divide an In Memory Column Store between the Primary and Standby • That is allowed between RAC Instances of the Primary OR the Standby for a given column store
45
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Analy?c queries on Ac?ve Data Guard
• Analy?cs queries used to fail when executed on Ac?ve Data Guard – They create temporary ADT objects during execu?on
• In 12.2, query on Ac?ve Data Guard will: • Create any needed ADTs on Primary , wait for ADTs to be replicated and applied to the Ac?ve Data Guard standby
• Proceed with execu?on u?lizing the replicated ADTs • Eliminates ORA-‐16000 (read-‐only database error) in this situa?on!
• Backport available for 11.2.0.4
46
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Diagnos?cs and Tuning for Ac?ve Data Guard
• Support for Diagnos?c Pack (AWR)
• Support for Tuning Pack features and SQL Plan Analyzer
47
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
AWR Support for Remote Snapshots and ADG (DB12.2) • Problem
– AWR snapshots cannot be taken in a read-‐only standby environment – Performance monitoring and analysis is limited to basic STATSPACK func?onality
• Solu?on – In Oracle Database 12.2, AWR framework enhanced to support capture of remote snapshots from any generic database including Ac?ve Data Guard (ADG) databases. • A target catalog database collects snapshots from the remote databases (sources) • Snapshots can be collected automa?cally or manually • AWR tables on the catalog database accumulate snapshot data from all sources via database links • Source databases must be registered on the catalog via new DBMS_WORKLOAD_REPOSITORY.REGISTER_REMOTE_DATABASE API
48
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
AWR Unified Management Framework topology
49
Source databases
AcCve Data Guard
Remote AWR Data
Store
Catalog of registered remote databases
Remote database could be a Primary and its standbys
or any other generic database
DBID1 AWR Data
DBID2 AWR Data
ADG
DB5
Primary
db_link_to_sources
db_link_to_target
DB1
DB3 DB4 DB2
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Remote Snapshot Configura?on for ADG Database
50
Primary
Local AWR Data
Catalog database
Remote AWR Data
ADG Reader Farm
Source databases
AcCve Data Guard 3. Define the source databases
4. Define the target database
5. Create the topology
6. Register the source databases
7. Register the source databases for AWR service
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Remote Snapshots: Features • Automated snapshots triggered from the catalog database • Control over snapshot interval and reten?on period per database • Container database support • Create snapshots manually on demand from the catalog database
– dbms_workload_repository.create_remote_snapshot(''TYPICAL'', sourceid); – Generate AWR report from catalog by running: @?/rdbms/admin/awrrp? .sql
51
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
SQL Tuning Advisor Support for Ac?ve Data Guard • Problem
– ADG databases are widely used to offload repor?ng or ad hoc query-‐only jobs from primary
– Repor?ng workload profile is different from primary and oVen requires tuning
• Solu?on • Oracle Database 12.2 introduces the ability to tune SQLs workloads running on ADG database • All SQL Tuning Advisor tasks issued at the standby
– Create tuning task, execute tuning task and implement SQL Tuning Advisor recommenda?ons – Test execu?on (heavy liVing) happens on standby, only minimal write related ac?vity on primary
• The required data for the above tasks are fetched from primary over a database link from standby • Task details and tuning results are stored at primary and the essen?al data required to construct the report is accessed remotely from primary
• The report is constructed locally at the standby, with no CPU overhead in primary
52
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
SQL Tuning Advisor Support for Ac?ve Data Guard
Primary Standby
dblink_to_primary
create tuning task execute tuning task report tuning task implement recommendaCons
– All changes are done on primary and propagated from primary to standby by redo apply – The data required for running the tuning tasks are fetched from the primary – Support for PDB level tuning – Test execu?on (heavy liVing) happens on standby, only minimal write related ac?vity on primary
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. | 54
Create SQL Tuning Advisor Support for Ac?ve Data Guard
Primary Standby
dblink_to_primary 1a. create_tuning_task : create a tuning task
1b. Write data about the task
AcCve Data Guard
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. | 55
Execute SQL Tuning Advisor Support for Ac?ve Data Guard
Primary Standby
dblink_to_primary
2a. execute_tuning_task : Fetch data from primary to execute a task 2b. Execute tuning analysis process
2c. Write tuning results to disk
AcCve Data Guard
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. | 56
Report then Implement the RecommendaCons SQL Tuning Advisor Support for Ac?ve Data Guard
Primary Standby
dblink_to_primary
3a. report_tuning_task: Fetch data from primary to produce and create a tuning report
3c. Write accepted profile results to disk
3b. Execute: accept_sql_profile
3d. Accepted profile available via redo apply
AcCve Data Guard
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Oracle Data Guard
• Session Draining support with the Broker • Preserving Applica?on Connec?ons to An Ac?ve Data Guard Standby During Role Changes
57
Transparent Client ConnecCon Failover
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Session Draining support with the Broker
• Session Draining works with the Oracle Drivers today in both Oracle Database 11g Release 2 and Oracle Database 12c Release 1. – But it is a manual process
• srvctl stop services on primary site & allow drain • Data Guard switchover (Broker or SQL*Plus) • New primary database open, srvctl start service, rebalance
• In Oracle Database 12c Release 2 the Data Guard Broker will take over the session draining so the above becomes one command.
58
DGMGRL> switchover to BOS1 wait;
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Preserving Ac?ve Data Guard Applica?on Connec?ons
• Preserve sessions on Ac?ve Data Guard during failover/switchover
• Drama?cally improve the user experience – Users stay connected, no disrup?on – Reduces concern for reconnect storm aVer failover
• Automa?c, nothing to set up. – Services need to be configured correctly
59
Read/Write Read
Read/Write
AcCve Data Guard Standby
Primary Failed Primary
Primary
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Protec?on ProtecCng your data to the max!
60
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Increase your Data Protec?on with Data Guard Release 12.2 • Automa?c Password File management • Easy repair of No Logging opera?ons • Data File Block Comparison between Primary and Standby • Mul?ple FSFO Targets and Observers • FSFO in Maximum Protec?on mode • Zero Data Loss Failovers in any mode with Storage Failures • More protec?on with Auto Block repair enhancements • Enhanced control of Alternate des?na?ons • Convert to TDE easily with likle or no down?me
61
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Data Guard and No Force Logging
• No-‐Logging opera?ons only generate enough redo to tell the standby that some blocks are now unrecoverable. – Repairing them meant restoring the complete affected data files.
• No-‐Logging blocks will be tracked by the Primary database • Standbys can then be easily repaired using RMAN
– Valida?on and repair only of blocks on standby that were invalidated by non-‐logged opera?ons (such as direct loads) on the primary
– This will query the Primary for block loca?ons that contain non-‐logged blocks and repair the standby database.
62
rman validate/recover .. nonlogged block
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Oracle Data Guard Database Compare
• High speed database compare between primary and standby – Detects latent corrup?ons on blocks that have not been touched by Oracle
• Useful in detec?ng latent corrup?ons, lost writes • Fast compare at physical level, accounts for blocks being modified • Network & compute efficiency
– Send & compare blocks in “batches” – Op?mizes for blocks where versions are the same – Ignores empty blocks
63
DGMGRL> validate database bos1 datafile 10 output=checkBOS1; Operation requires a connection to database "bos1" Connecting ... Output files are created in /…/BOS1/trace on host “Boston1"
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. | 64
Oracle Data Guard Database Compare Summary: *************************************************************************** ID: Block Type Id TOTAL: Total number of blocks found DIFFV: Number of block pairs with different version LWLOC: Lost Writes at Local LWRMT: Lost Writes at Remote SAMEV: Number of block pairs with same version SAMEV&C: Number of block pairs with same version and checksum DIFFPAIR: Number of block pairs with same version but different contents ENCERR: Undecided blocks related to encryption/decrytion error. e.g. Wallet is not open. SKIPPED: Skipped blocks due to data corruption, etc ID TOTAL DIFFV LWLOC LWRMT SAMEV SAMEV&C DIFFPAIR ENCERR SKIPPED 29 0000001 0000000 0000000 0000000 0000001 0000001 0000000 0000000 0000000 30 0000125 0000000 0000000 0000000 0000125 0000125 0000000 0000000 0000000 58 0000512 0000512 0000000 0000000 0000000 0000000 0000000 0000000 0000000
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
TDE Tablespace Conversion
• Requires an Export and Import to encrypt • Cannot be done online so data is unavailable during the process • Outage for the data can be mi?gated by using a Transient Logical Standby
– Convert a Physical Standby to a Transient Logical Standby • Export the data, Create the encrypted tablespace, Import the data at the transient logical standby
– Switchover – At hkp://www.oracle.com/goto/maa see these two papers
• Conver?ng to Transparent Data Encryp?on Using Ac?ve Data Guard 12c (DBMS_ROLLING) • Conver?ng to Transparent Data Encryp?on Using Data Guard Transient Logical Standby with Oracle Database 11g
65
Today
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
TDE Tablespace Conversion
• New simple SQL command to encrypt data files in place
• Data is s?ll unavailable during conversion • Outage for the data can be mi?gated by using a Physical Standby!
– Stop Redo Apply, execute the command, Restart Redo Apply • To encrypt SYSTEM and SYSAUX the standby must also be in MOUNT mode
– Switchover – Repeat steps on new Physical Standby (Original Primary)
• Backport available for 11.2 and 12.1 66
Tomorrow
alter database datafile '+DATA/BOS/DATAFILE/users01.dbf' encrypt;
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
TDE Tablespace Conversion
• Tomorrow’s (çprevious slide) offline data file method s?ll available – Complete with Physical Standby method
• A new SQL command will arrive on the scene!
• Encrypts all data files in the tablespace online. – Primary only & Requires auxiliary disk space
• Can also be done OFFLINE for beker performance – No Extra storage, but requires data outage on Primary. – Offline can be done on a Physical Standby first as with the previous example
67
Future (12.2 only)
alter tablespace users encryption encrypt
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
A look at another customer Implemented Oracle Data Guard Far Sync to achieve Zero Data Loss over 1340+ Miles
68
Oracle Active data Guard 12C
Mr. Alok Khanna – ED IS Mr. Deepak Agarwal – GM COIS Mr. Rajan Wahal – CM COIS Mr. Narinder Kumar – CM COIS Mr. Manish Srivastava – Sr Mgr COIS Mr. Vikas Srivastava – Sr Mgr COIS Mr. Lokesh Rana – Mgr COIS Corporate Information Systems, IOCL,Gurgaon
Business Continuity Architecture (Previous)
Standby Database Server(Bangalore)
SAP Application
Servers
Primary Database Server(CBTC)
DB
Managed Recovery
Archived Logs
DB
Archived Logs
Online Logs
DB
Archived Logs
Standby Redo Logs
Standby Database Server(NRS)
WAN File Based Log Shipping
DataGuard Services
Corporate Information Systems, IOCL,Gurgaon
Current Architecture
ORACLE ACTIVE DATA GUARD (FIBRE CHANNEL; BW=60Mbps)
OLTP Database Size 9.13TB
Redo generation Rate
6.24MB/s
RedoLog Size*Number of redo logs
2GB*8
Redo Transport Compression ratio
75-80%
DB_LOST_WRITE_PROTECT
FULL
Corporate Information Systems, IOCL,Gurgaon
Oracle ADG Solution
Corporate Information Systems, IOCL,Gurgaon
Fast Sync Async
Priority Route Alternate Route
Primary
Far Sync A
Standby Database
Far Sync B
CDC, Gurgaon NRS, Gurgaon RBCSC, Bengaluru
Switchover
Priority Route Alternate Route
Async
Far Sync A
Far Sync B
Far Sync A is active
Max Availability
Corporate Information Systems, IOCL,Gurgaon
Benefits Obtained
Corporate Information Systems, IOCL,Gurgaon
v Near zero lag between Primary and Remote site separated by 2000 kms
v Best Data Protection v Automatic Block Repair – High Availability v Redo transport Compression v Simplification of BCP Process
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
• Ease of use – Use DBCA and EMCLI to create Standbys – Use RMAN and Enterprise Manager to create Far Sync
Instances in addi?on to Standbys – Rest Interface to the Data Guard Broker – Chef tools to create and manage Data Guard standbys – Puang the FSFO Observer in the Background – Scrip?ng with the Broker DGMGRL command line interface
• Speed up Performance and Role Changes – Use the In Memory Column Store on your Standby – Use AWR and SQL Tuning Advisor to diagnose and tune
Redo Apply and Query issues on your Ac?ve Data Guard Standby
– Move users transparently during role change with Session Draining on the Primary and Preserving exis?ng connec?ons at the Ac?ve Data Guard Standby
76
• More power to use Data Guard – Using all your Standby Instance to apply redo – Easily crea?ng Subset Standby databases in Mul?tenant – Data Guard Broker Support and involvement in
DBMS_ROLLING controlled rolling upgrades – Migrate and Failover PDBs with the Broker
• Increase your Data Protec?on – Automa?c Password File management – Easy repair of No Logging opera?ons – Data File Block Comparison between Primary and Standby – Mul?ple FSFO Targets and Observers – FSFO in Maximum Protec?on mode – Zero Data Loss Failovers in any mode with Storage Failures – More protec?on with Auto Block repair enhancements – Enhanced control of Alternate des?na?ons – Convert to TDE easily with likle or no down?me
Data Guard Release 12.2
Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
Safe Harbor Statement The preceding is intended to outline our general product direc?on. It is intended for informa?on purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or func?onality, and should not be relied upon in making purchasing decisions. The development, release, and ?ming of any features or func?onality described for Oracle’s products remains at the sole discre?on of Oracle.
78
Top Related