Back with the latest update.[ the day has been crazy already and its not done yet. I am finally finished my presentation “An ETL Framework using OWB”. I think it went well. It was good to see so many people who are interested in this compelling technology. The basic message I had for people is that for ETL…..keep it simple. And as for the choice of OWB; it is one that your budget will like and your developers will be able to use to produce most of the results that you need. It is a compelling ETL choice. Thanks to everyone who attended and took part in the session.
So now on the next and related session, the Oracle 11g Database for Data Warehouses with Ray Roccaforte who is one of Oracle leading data warehousing gurus. It started with a discussion of the latest version of Database Machine (Exadata), he explained how they were able to achieve a huge performance gain over the orginal version of the Exadata server. With improved everything including CPU, interconnect, memory and so much more, has provided a doubling of the performance. As well the latest compression and new storage indexes which reduces I/O to near zero., this is done by ensuring that the min and max values are within a read range, so it only read what it needs to. In addition the new columnar compression allows for more effective storage. It stores data based on a column and then compressed….very cool. No quite a column database but they seem to be getting closer.
The discussion around the features in teh database is one that I always like. So consider that OLAP and Data Mining and Analytics are built right into the database and can access data effectively and efficiently. The tighter integration in 11g with OBIEE and the database is another encouraging feature. These are areas that poeple may not be aware that they exist, so I would encourage people to take a look at this technology it could save real dollars considering the cost of other analytic software today.
So the message is that Exadata is a real option for DW deployments and they are pushing this message out to the audience today
Ian I attended your session at Open world 2009.It was grt. I hade some question if you could help be understand how could I Identify the CHANGED rows from the source system before loading in DW.(Identify new records is simple)I remember you saying some thing about Hash(#) values comparison. Can you provide me some more info on it. I use OWB and have problems identifying changed data to load into dimensions and facts tables. As a result right my DW cannot support Point In Time reporting data and looks like latest snap short of production system.
ReplyDeletemy mail id is ashwinponda@gmail.com
guys what you can do is create a md95 hask key as a colum and insert all the velues that are not in the target.
ReplyDelete