Today the RittmanMead BI Forum 2013 started with an Oracle Data Integrator Master Class provided by (in alphabetical order) Mark Rittman, Michael Rainey and Stewart Bryson.
I made some notes, so I’ll try to highlight some points from the Master Class. It won’t be complete, but it gives an indication. I didn’t know much about ODI, so the start of the Master Class was very welcome to me.
Overview by Stewart Bryson
Just in time for Oracle BIA 11.1.1.7 – everything ODI, bye, bye Informatica and DAC (DAC translated into configuration and packaging and load plans).
Further integration with Oracle Weblogic and the Fusion Middleware
ODI & OBI Jive really good together. Abstraction of physical Sources logic built into the model not the process. Same like OBI. Develop the model, not the query. Metadata. See the the similarity between ODI and OBI.
Scripting becomes Groovy with ODI (see later on).
Everything you do in ODI goes via Agents
OCI versus JDBC –> ETL versus E-LT (insert into select)
Develop once in the Model instead of in the interface / mapping. Interface inherits from the model, override if needed. Interconnect interfaces via yellow temporary interfaces
Knowledge Modules (KM) – the physical implementation of the logic
How to load is determined by the KM, E-LT or ETL
Packaging (sequencing) versus Load Plans
Goldengate and ODI by Michael Rainey
Oracle Reference Architecture for Information Management and Big Data – 2013 Whitepaper
Oracle Goldengate – Data Replication
CDC Journalizing in ODI – Journalizing Knowledge Module (JKM)
Goldengate uses the JKM….generated metadata and a readme how to apply. No full integration
RMAN redo log, archive log , cleaning the logs
Parent Child Journalizing two different tables via two interfaces into the same target table
Subscription Views a RM feature to provide a choice for the ETL developer to choose which object to use
Oracle Goldengate and ODI are a Perfect Match
ODI and Big Data by Mark Rittman
If you do not care about a single row you refer to Big Data
Decisions based on your own data vs decisions based on all data relevant to you
Oracle Big Data
Terminology
– HDFS
– Hive
ODI (or OBIEE) to connect the Hadoop world to the traditional DWH world via the Hive.
ODI and OBIEE als enabelers to use the result out of the Hadoop world
Check Mark’s presentation
The three R’s of ODI – Resuming, Restarting, Restoring by Stewart Bryson
Using database features in ODI to eliminate the “Aftermath scenarios”.
Resuming
Enable resumability in ODI to prevent a load to fail and just suspend
Monitor the resumable view – dba_resumable view
Load Plans
Restarting
Flashback Table
Flashback Database
Record the SCN (be careful with row level and block level)
Restoring
archivelog mode
Block change tracking file
Insert /*APPEND*/
Nologging
Backup
Check the origin of the “Three R’s of Data Warehouse Fault Tolerance” here.
Automating & Scripting Oracle Data Integrator by Michael Rainey
Some great examples of how to make life a little bit easier. Could Oracle have used this to migrate Oracle BIA from Informatica to Oracle Data Integrator earlier?
ODI – 11g Expert Accelerator for Model Creation – David Allen
All in all it was a very valuable Master Class. I learned a lot and there is lot more to investigate and explore. Hopefully this is the standard for the rest of the BI Forum.
One thought on “RM BI Forum 2013 Notes – Oracle Data Integrator Master Class”