What are "Queries from Hell"?
Posted at 2/20/2008 04:24:00 PM
In the world of data warehousing, administrators need to be wary of "queries from hell." These are variously defined, by IT Jungle as "gigantic queries that may have been accidentally set off by users and that can consume enormous amounts of systems resources," and by the IS Management Handbook as database queries that "can destroy perceived levels of performance if they are not identified and managed carefully."
They are defined here simply as queries that are really really hard to execute using standard data warehouse technology because they require database restructuring, days of SQL coding, inordinate processing resources, or just don't seem to adhere to the rules that 90+% of "standard" queries follow.
Telling business users that they can't have an answer to a question because even though the answer lies within a database, it is simply too difficult to extract, is not a feasible option. Neither, often, is explaining that the answer can be obtained, but it will take weeks or months of programming.
The answer is to segregate these queries from the 90+% of standard BI and business analysis queries, and treat them differently. Use the right tools for the job. That's the kind of thing you'll be seeing in this blog over the coming months and years.
They are defined here simply as queries that are really really hard to execute using standard data warehouse technology because they require database restructuring, days of SQL coding, inordinate processing resources, or just don't seem to adhere to the rules that 90+% of "standard" queries follow.
Telling business users that they can't have an answer to a question because even though the answer lies within a database, it is simply too difficult to extract, is not a feasible option. Neither, often, is explaining that the answer can be obtained, but it will take weeks or months of programming.
The answer is to segregate these queries from the 90+% of standard BI and business analysis queries, and treat them differently. Use the right tools for the job. That's the kind of thing you'll be seeing in this blog over the coming months and years.
