Edgewater Ranzal presented at ODTUG Kaleidoscope (KSCOPE) 2015 about implementing a fully-functional DR system leveraging Oracle Enterprise Performance Management (EPM) and Oracle Business Intelligence (BI).
3. Business Intelligence
• Scorecards & Dashboards; Mobile Delivery
• Prepackaged Analytic Applications (OBIA)
• BI Foundation (OBIEE / Essbase)
• Financial & Operational Analytics
• Reporting & Publishing
• Visual Exploration
• Information Discovery (Endeca)
• Business Intelligence Cloud Service (BICS)
Financial Close & Consolidations
• Legal, Segment, & Mgmt Reporting
• Financial Performance & Close
• Optimization & Performance Lab
• Tax Provisioning
• Regulatory Requirements
Data Integration Services
• ERP Integration
• Financial Data Management
• Data Warehousing
• MDM & DRM
• ETL Services and Automation
Infrastructure
• Hardware Sizing, Installation, & Configuration
• Exalytics Benchmarking
• System Monitoring, Backup, & Disaster Recovery
• Upgrade & Migrations
Enterprise Planning & Analytics
• Strategic Planning (HSF) &
Modeling
• Planning, Budgeting & Forecasting
• Workforce, Capital, & Project
Financial Planning
• Public Sector Planning
• Cloud Services
Costing & Profitability Management
• HPCM Standard & Detail Models
• Waterfall Allocations
• Shared Service Charges
• Activity Based Costing
• Customer, Product, Segment &
LOB Profitability
Advisory Services
• Project/Program Management &
Process Change
• EPM & BI Road Maps
Training & Support Services
• Standard & Custom Training
• Virtual Classroom
• Multimedia and UPK
• Service Level Agreements
• Managed Services & Hosting
• Performance Engineering
Financial Close &
Consolidations
Business
Intelligence
Enterprise
Planning &
Analytics
Data Integration
Services
Training &
Support Services
Advisory ServicesInfrastructure
Costing &
Profitability
Mgmt
Our Services
4. The Sherwin-Williams Company
The Sherwin-Williams Company was founded by
Henry Sherwin and Edward Williams in 1866.
Today, we are global leader in the manufacture,
development, distribution and sale of coatings and related
products to professional, industrial, commercial and retail
customers.
The Company manufactures products under well-known brands such as
Sherwin-Williams®, Dutch Boy®, Krylon®, Minwax®, Thompson’s® Water
Seal® and many more. With global headquarters in Cleveland, Ohio, Sherwin-
Williams® branded products are sold exclusively through more than 4,000
company-operated stores and facilities, while the Company’s other brands are
sold through leading mass merchandisers, home centers, hardware stores,
automotive retailers and industrial distributors.
www.sherwin-williams.com
5. The Sherwin-Williams Company
Largest Producer of Paint & Coatings in US; among Top 3 worldwide
Founded in 1866 in Cleveland, OH
2016 marks 150th Anniversary
2013: $10.19 billion in sales
Business in 120+ Countries
34,000+ employees
4,000+ Company-Owned Stores
90+ Manufacturing Facilities Globally
6. A fully functional DR system
● EPMA Deployments
● Essbase Studio deployments
● Planning
● Essbase
● FRS
RTO – 8 hours
RPO – 24 Hours
Use existing infrastructure
No VMs
The Challenge
7. Backups
● Obviously to support Production issues
● To keep DR application in sync
● Failsafe when switching to DR
● Need
● RDBMS
● Rman and datapump export
● Essbase
● Export (level 0 or full of all apps)
● Application metadata
● Full ARBORPATH and MIDDLEWARE_HOME (in case of media failure)
● LCM
● EPMA
● Great for one off artifacts
● Good to have after DR failover for insurance
Important Prerequisites
9. If you want to connect to Oracle RAC via infiniband, you
cannot use OVM
You want to save the resource overhead required by the
OVM hypervisor
Prevent oversubscription and performance issues
associated with OVM
More consistent disk and network I/O performance
It is an extra skill set and technology that needs to be
accounted for
One less point of failure
Allow more direct control over the environment (i.e. DR
Process)
Why use bare metal over Oracle VM?
10. Here are the standard ports and what we set for the
ports for the 2nd instance:
How to change the ports for the 2nd instance
11. Most of the ports are weblogic-based and can be changed in the EPM config tool or
in the weblogic admin console
R & A RMI ports are changed via the config tool
Planning RMI port is tricky; it needs to be changed in the EPM registry and in the
following files:
/oracle/Middleware/EPMSystem11R1/common/RMI/11.1.2.0/HyperionRMIRegistry
And in this file as well :
/oracle/Middleware/user_projects/epmsystem1/RMI/HyperionRMIRegistry
Essbase studio ports are changed via the EPM registry
OPMN ports are changed via the following file:
How to change the ports
12. Essbase Studio & EPMA
Create EPM REG scripts to change hostname &
ports as part of the failover process
How to handle products that are not HA
capable to work in an automated DR fashion
15. How to Handle the RDBMS Oracle Data Guard
Oracle Data Guard ensures high availability, data protection, and disaster recovery
for the Oracle DBMS through the creation and maintenance of a standby database
which is a copy of the production database. If the production database becomes
unavailable because of a planned or an unplanned outage, Oracle Data Guard can
switch any standby database to the production role.
16. Oracle Data Guard Setup
1. Configure all EPM Products to use the Data Guard URL
1. Allows transparent connection to the Standby in a Fail over situation
2. JDBC URL for Hyperion when using Oracle DataGuard (Doc ID 1359130.1)
3. Bug 20729245 - ESSBASE DOES NOT START WITH DATA GUARD URL
2. Create a Physical Standby
1. Provides a physically identical copy of the primary database, with on disk database structures that are identical to the
primary database
2. A physical standby database is kept synchronized with the primary database, through Redo Apply, which recovers the
redo data received from the primary database and applies the redo to the physical standby database.
1. http://docs.oracle.com/database/121/SBYDB/create_ps.htm#SBYDB00200
3. If using 12c apply this patch https://updates.oracle.com/Orion/Services/download?type=readme&aru=17704007
1. https://support.oracle.com/epmos/faces/BugDisplay?_afrLoop=3939865576206499&id=17877323&_afrWindow
Mode=0&_adf.ctrl-state=obgnx2qnt_224
1. ORA-600 [KRBI_INIT_FRA_METADATA_NOT_BACKUP_CTL] RMAN DUPLICATE FROM ACTIVE
DATABASE with STANDBY as TARGET (Doc ID 1618470.1)
3. Create a Data Guard Broker Configuration
1. Provides a management system to maintain and monitor the configuration and interactions between the primary
database and the standby
4. Monitor
1. We use Grid control to alert us when there are issues with the standby.
18. Oracle Data Guard Switchover
We switchover to the standby database instance using
Data Guard
● The instances switch roles in which the DR database
becomes the primary and updates the standby database
(which was PROD)
● In a true DR situation, this would be done as a failover
because the PROD database would not be available
● Use dgmgrl
19. Oracle Data Guard Switchover
● In our case, we have to manually mount the standby
instance; this is due to a bug with the local_listener
parameter in the database which needs to be fixed
● srvctl start database -d <dbname> -o mount
● Once that is done, archive log shipping resumes to
PROD (which is the standby now)
20. 1. Nightly backups that:
a) Back up essbase data
b) Back up essbase artifacts
c) Back up outline
2. Copy the files to the DR server using rsync
3. Sync the RM1/LCM directories using rsync
How to setup Essbase replication scripting
21. 1. Restore the backups to the essbase server
a) If a new database/application create it
b) Copy outline to the application
c) Copy the artifacts
d) Load data
22. We are using an rsync process to sync up a
backup area and the RM1/LCM directories
rsync -avz source directory>
user@server:<target directory>
A replicated NAS (such as NetApp) could use
be used instead of rsync
How to configure replication of EPM shared
directories
23. Create a friendly name in DNS like
“hyperion.company.com”
This can then be controlled via changing the DNS
pointer or by pointing it to a VIP in a load balancer
When the environment fails over to DR, it will be
transparent to the end users
How to setup a dynamic host naming
convention for failover & failback
24. We got it down to 8 easy steps!
1. Run stop.sh/.bat on Exa01, Exa02
2. Fail Data Guard instance to DR
3. Execute “sq_script_prod_to_DR.sql.bat” (changes Planning data
sources for Oracle & Essbase Host names)
4. Execute “switch_to_dr.bat” (makes necessary EPM registry changes for
non HA aware products like EPMA & Essbase Studio)
5. Flip Rsync Job from PROD -> DR to DR -> PROD
6. Fail Data Guard back to PROD
7. Run start.sh/.bat Exa01, Exa02
8. Manually change FR data source properties in EPM workspace (this is
stored in a binary file in the RM1 folder and thus cannot be scripted)
Failover & Failback Process
25. We had to manually reconfigure Shared Services
connection using a registry edit command on
failover and back
● In versions past, you can use a Data Guard URL
● JDBC URL for Hyperion when using Oracle Data Guard (Doc ID
1359130.1)
● This does not work with a RAC database and Data Guard and
the note above needs to be updated
Challenges Encountered
26. ● Readme for 11.1.2.3.0 states:
● 16618139, 16515442 -- High Availability. When the Oracle
Hyperion Foundation Services database is configured on an
Oracle Real Application Clusters (RAC) database in which the
JDBC URL connection string specifies more than one node of the
RAC database, Essbase fails to start in failover mode.
● The SCAN listener work around also does not work if you want to
include the Data Guard instance in the URL also, even though the
note states you cannot configure 2 nodes in the URL (which is
why you use SCAN), it also seems to apply to 2 different
databases (which was allowed in prior releases)
Challenges Encountered
27. ● Readme for 11.1.2.3.0 states:
● This will not work
● jdbc:oracle:thin:@(DESCRIPTION=(CONNECT_TIMEOUT=3)(AD
DRESS_LIST=(FAILOVER=on)(ADDRESS=(PROTOCOL=TCP)(H
OST=PROD-
SCAN)(PORT=1521))(ADDRESS=(PROTOCOL=TCP)(HOST=DR-
SCAN)(PORT=1521)))(CONNECT_DATA=(SERVICE_NAME=EP
M)))
Challenges Encountered
28. ● Issues making the LWAs work in DR
● Had to update some ports not to use LWA for Calc
manager, Planning
● EPMA Planning Application Deployment Hangs at 10%
Without Errors (Doc ID 1334738.1)
● The hardware load balancer has not been configured to
accept loopback requests that come from the internal
Oracle processes
● Issues making Essbase Studio start
● startServer.sh & stopServer.sh scripts did not have
execute permissions; once I chmoded them, they
worked. This appears to be a bug with the Linux
installer/configurator
Challenges Encountered
29. ● Issues making Essbase Studio start
● Need to modify
$EPM_ORACLE_INSTANCE/config/starter/Essba
seStudio.properties
● > start.script/BPMS/bin/startServer.sh
● env=JAVA_HOME=/u01/app/oracle/product/EPM/Middl
eware/jdk160_35/jre,JAVA_OPTIONS=-
DESSBASE_STUDIO_INSTANCE=/u01/app/oracle/pr
oduct/EPM/Middleware/user_projects/epmsystem1/BP
MS –DsuppressAPSProductInf
● stop.script/BPMS/bin/stopServer.sh
Challenges Encountered
30. ● Issues making Essbase Studio start
● added the following highlighted lines to the
“server.properties” on xlytwv01; this caused Essbase
Studio to override what is the EPM registry.
● transport.port=5410
● catalog.url=oracle://scan-listener:1521/SERVICE
● catalog.username=studio_schema
● catalog.password=<encrypted value>
● url=http://host:12190/eas
● server.httpPort=12190
● Make sure to not use
server.enableNativeAdministrator=true in
server.properties; if you do, you won’t be able to login
to the Essbase Studio Console.
Challenges Encountered
31. ● Updating Hyperion Shared Services (HSS)
Registry with Host Name for
SHARED_SERVICES_PRODUCT/LOGICAL
_WEB_APP Fails with Error "Multiple host
components exist for host name" (Doc ID
825974.1)
Challenges Encountered
32. ● RMI Port issue
● check the HyperionRMI_Port.properties and the
registry; both are set to the custom RMI port of
11443
● Also check:
$EPM_ORACLE_HOME/common/RMI/11.1.2.0/Hyp
erionRMIRegistry
● Change port there also
● F5 LWA port and port 10080 must be opened to
use LWA for EAS
● F5 timeout settings
Challenges Encountered
33. ● Essbase subvars need to be loaded manually
● This can be done by using the LCM backups from
production that are carried over to DR
● Manually load through EAS export/import
● EPMA Applications can be deployed to DR
System if needed after failover
● This can account for any inconsistencies in RDBMS to
Essbase backup timings
● Aggregate Data
● Test
Some things that may have to be manually
done after failover