Quote:
Originally posted by slimdave
Well, I'm going to have to go ahead and disagree with you there.
1. It tells you that it's an academic question not a real life one. 5 million rows, but only 100 new rows per month? Gimme a break.
2. Rows-shmows -- it's blocks that count. If the rows are distributed evenly then that's probably 100% block retrieval, so a full-scan would be faster.
3. Update it once per year per row, which would meet a definition of infrequent, and you've still got 5 million changes per year. Assume they only change during working hours and that gives you 40 changes per minute.
4. Yes, but that's not the primary concern, just a nice feature.
5. Moot point, since the index would probably never be used, but b-tree's aren't that inefficient.
good answer..:-) is this the reason why you dont want to do ur