Now after some research - I detected one simple rule for provoking In-Memory scans :
Now finally after a long period of waiting - and looking presentations and reading blogs I was able to do In-Memory databasing on my own. Having sampled some theoretical background how a a pure In-Memory Db works with SAP HANA; my expectations on the Oracle In-Memory Option were high. Also because Oracle promised it would work without code change, out of the box.
The Oracle database is located on Sun Solaris Machine with enough memory and 32 Xeon-Cores; The Memory Pool is sized 64G, compared to the rest of the SGA , buffer cache , shared_pool ... 4G, and 300G of data.
Introduced as an Option Oracles In-Memory option will change the world of databasing also like SAP HANA does; Since July the release is out but the search for resources and documentation is poor;
Here some useful links I found.
First the Home of Oracle In-Memory
About 3 years ago SAP started to create a new database engine, SAP HANA, with a pure In-Memory concept. SAP aggressively move it's new database to it's software stack. The database was not used as cache but for running Enterprise Application satisfying both OLTP and OLAP demands on the same database.
As oracle announced the new Oracle 12c in late 2012 there the 'c' was for cloud based computing, means the simplification of creating several database instances under the hood of a container parent database. For the end user this wasn't a visible benefit. It seemed Oracle did ignore SAP HANA. But even you cannot compare number installed databases for SAP HANA and Oracle HANA made an impact. Something new and very visible to the end user arises.
Now about a half year Oracles In-Memory Option announcement the European launch event was done on Tuesday this week. In the radison Blue hotel - which was a very exciting and comfortable place for that - the conference room was much more filled up with the oracle followers then the soccer arenas of this years soccer world championship.
The event was well organized and mixed up with high professional speakers.
What still in my mind was ...
Maria Colgan introduced the more detailed usage and environment of the Oracle In-Memory Option. She did it in a so clear and compact form - I am really impressed. Also the life demos have been amazing.
I was also impressed about Dr. Maaike Limper's session. She works as scientist at CERN; and tested to use the In-Memory Option to get faster analysis about particle collisions and used a data model of particle typed tables with hundreds of parameter per particle and immense number of rows. By using the In-Memory Option she said it was possible to play with data to drill down and possible find something new due the detected data patterns of the sampled particle sensor data.
Finally Dr. Dietmar Neugebauer held a session like 'is the DWH dead now?' which proofed clearly the DWH is not only about analytic queries and so superfluous but also to consolidate and validate data from different data-sources/systems of the whole company. So the DWH is not dead with introduction of the new In-Memory Option. Maybe some 1:1 replication of operative data will get obsolete.
At the end of the event everybody knows and feels something has happened in the database world which will be visible for all end users and will have a tremendous effect on system landscapes and software development - back to database centric ultra-fast processing.
Matrix : What you must learn is that these rules are no different than rules of a computer system. Some of them can be bent, others can be broken. Understand?
-- SCOPE : OASIS
-- DDL:CALLITEMS_UK01 :INDEX.MOD - TEST
DROP INDEX CALLITEMS_UK01;
CREATE UNIQUE INDEX CALLITEMS_UK01
( CASE UNIQUE_REF WHEN 1 THEN UPPER(TRIM(ITEMCODE)) ELSE NULL END
, CASE UNIQUE_REF WHEN 1 THEN ITEMTYPE ELSE NULL END
, CASE UNIQUE_REF WHEN 1 THEN ACTIVESINCE ELSE NULL END )
Dependend on the uniqeness flag UNIQUE_REF a row may be unique to others or not. Maybe this makes sense for specific call item types. In our project a CALLITEMTYPES table controlled the unqueness of specifc call item types , populating the UNIQUE_REF flag to the CATLLITEMS table.