File Library

Have you ever wondered why some loads are fast and others are slow? Why the same query performs differently at different times? Why your queries and loads are not as fast as they should be? Then this is the session for you.

Starting with a discussion of how data file IO is handled in Windows and Unix, techniques to ensure apples-to-apples testing are presented. Then using the results of over three hundred load, and calculate/aggregate tests using very large BSO (9gB input level 175gB Calculated) and ASO cubes (1.4 billion cells 84GB aggregated). The testing spans Windows and UNIX; ASO and BSO; varying cache settings; sort order and file formats are presented. These variables are evaluated and ranked with several new and surprising conclusions. Conclusions, that in some cases, run contrary to existing best practices.
Finally, expanding on the chapter "How ASO Works and How to Design for Performance” in Developing Essbase Applications, the speaker will discuss surrogate keys, MDX, multi-attribute queries.
All of this will be discussed in light of real world experience where multiple cubes are running and data is prepared and hardware supplied by other parts of the organization with differing practices and priorities. In short: real techniques you can implement when you return from the conference.