Tony,

Awesome point that I think too few people consider (or even know to consider). I always think about this when creating a new table, but I don't always know exactly how that table will be used in the "wild". I've even re-clustered tables upon seeing how they're typically queried [sp?].

Even when I recluster, I recognize that there will be a price to pay with inserts. I personally haven't dug deeply enough to know when that price is worse than the SELECT improvements. I generally just consider the situation -- more INSERTs or more SELECTS?

Unfortunately, I work with a crew that feels that every table should have a clustered IDENTITY index that "resolves any abiguity" -- even though 90% of the time, the tail end of a relation will NEVER query that relational ID!

In general, our DB's are small enough that that is an OK approach. HOWEVER, we have one MASSIVE DB that has benefitted from some reclustering, and could benefit more, but there is reistance due to the above beliefs. We'll see how that sorts out over time, as we get to the 5-6 million-plus records (which is fast approaching).

-----------

Wish I had a clever tagline here... D.

Jsut thinking -- on the tagline -- what do you all think of "Inspiriation / Perspiration -- smell is in the nose of the beholder" or something like that? Thanks! D.