This week I am attending the RMOUG 2-day conference in Denver. This is one of the biggest (if not the biggest?) user groups, and this shows in the number of people attending the conference. Speaking to the organizing team yesterday on day 1, they said there were 960 registrations and countless walk-ins from members who work locally in the Denver area. Which makes this a great place for Oracle product managers to come and speak. There are a number of tracks
running across both days:
- Database Administration
- Data Warehousing
- Application Design
- Technology Management
- Web Internet/Intranet
The quality of the presentations is exceptionally high and all the presentations are about real-life experiences gained on various projects and/or engagements. Today I attended some extremely interesting sessions on data warehouse design, change data capture, overview of data warehousing features and an introduction to Warehouse builder. All were very well attended and very well presented.
From a data warehouse design perspective the paper on EAV (Entity Attribute Value) data modeling caught my attention. I have used this technique many times as an OLAP consultant and seen other OLAP consultants use it when dealing with highly attribute driven data and the need to allow users to analyze data across any number of different attributes. However, this paper provided a formal framework for implementing this type of design which way beyond my modest efforts when I was in consulting. The presenter, Ralph Hughes from Ceregenics, was excellent. He had a fully worked example in his presentation that showed how to implement EAV using relational tables, allowing developers to create a data warehouse within weeks without having access to the complete data model or exact details of the types of attributes that need to be tracked. It would be interesting to try and implement some of this back at the office and see how it works, and more importantly, try applying to an OLAP data model.
Poor Man's Change Data Capture (CDC) was a great presentation on how to manage data changes in a data warehouse where you don't have the benefit of a formal CDC mechanism. Kent Graziano, who is well known and highly respected member of the ODTUG community, explained how to use the DECODE function to manage CDC in a very intelligent way. Again, code snippets were included in the presentation, which helped to explain how this would be implemented. This was another example of something you could just take straight back to the office and implement right away. These are the types of presentations that make conferences like this so successful. The pay back is immediate.
On my part, I presented a paper of modeling multi-dimensional data within Warehouse Builder. The aim was to show that moving forward and extending a data warehouse to use multi-dimensional model does not require ETL developers to learn a whole new way of modeling. We have simply extended the relational dimensional model to include additional multi-dimensional metadata. This means you can now model a logical dimension quickly and easily and then at run-time decide to implement it as a relational table, a relational multi-dimensional table or a dimension within an anayltic workspace. This logical modeling allows ETL developers to use all the normal data sources, extensive transformation library and process flow events when creating a true multi-dimensional implementation. This will move OLAP into the main stream as now there is no reason to not include OLAP as part of your data warehouse model.
The final session I attended was by Jon Arnold on Warehouse Builder. This was a great introduction to the product. Jon has a lot of experience of data modeling and building data warehouses and this certainly came out during the presentation. He used a lot of examples based on his work at Denver Public Schools. Which again goes back to the heart of this conference. Real people talking about real world implementations. Which is what everyone wants to hear, right? If the session library of white papers and presentations are hosted on the RMOUG website I would certainly encourage everyone to download them as each one is packed with useful information.
Day 2 is just starting so I will provide more feedback tomorrow on the events and session from day 2.