Debating

May 3rd, 2006 by Peter Scott

Earlier this year Mark Rittman and I briefly discussed the concept of a matching pair of conference papers on aggregation in Business Intelligence. Mark would look at the use of Oracle OLAP as an aggregation technique and I would do something similar on conventional (and maybe less conventional) relational techniques. Both papers would be standalone, but the themes would complement each other. From Mark’s Blog I see that he has submitted an abstract for the UKOUG November conference on Analytic Workspaces. Well not to be outdone, my abstract on a comparison of Group By, Grouping Sets and Group By Rollup went in last week. It is now down to our peers to decide if we will go head-to-head (or is it head-to-tail?)

Some of you may know my views of staged debates at conferences; I just do not like them. How can I take a side in a debate on ‘the best design’ for a data warehouse or whether to use ‘surrogate or natural keys’ when I know in my heart that the true answer is ‘it depends’. Such debates usually come down to the characters of the debaters and usually degenerate in vitriol that serves no good. But the idea of contrasting papers holds water; delegates can take the content away and make their own assessments, the presenters can argue both the positives and the negatives of their own cases; in short, a better quality of debate by not debating.

Posted in Conferences, Data Warehousing, Oracle OLAP |

Comments

  1. Mark Says:

    May 3rd, 2006 at 10:40 am

    Hi Pete

    Very interesting. I did want to cover off GROUP BY … ROLLUP in my paper as well, but I don’t think I’ll do it as much justice as you will. It’s a shame we work for competitors as it would make a lot of sense to submit a joint paper, with the same dataset. Perhaps we can at least co-ordinate what we’re doing, at least use the same dataset (I was thinking about the SH schema) so that comparisons can be made.

    Funny enough, I think that this is a bit of a hot topic within Oracle. My spies tell me that AWs and MVs will be a lot more integrated in 11g, so I think any research we do will pay off in the future.

    regards

    Mark

  2. Pete_S Says:

    May 3rd, 2006 at 10:55 am

    Funny enough, I think that this is a bit of a hot topic within Oracle. My spies tell me that AWs and MVs will be a lot more integrated in 11g … and I have my spies too - might be the same ones ;)
    I agree, it’s a shame we work for comeptitors - a joint by-line would not work for my masters.