Archives of the TeradataForum

Message Posted: Wed, 14 Jan 2004 @ 16:25:08 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

I would like to further my understanding related to this particular issue --

We have a large table (millions of rows)

One column (not in the primary index) has the same value for all rows.

We do not collect statistics on the column currently

It takes over 6 hours to collect stats on this table so adding another column statistics collection would result in a longer run time

Joins/where clauses which use this particular column are coded in our front end user tool and cannot be removed quickly -- in reality, this column does not need to be referenced at all

Would it be beneficial to collect stats on this particular column or would the random amp sampling be sufficient to produce the optimal join plan. Again, ALL rows have the same value. I could understand collecting statistics if 2-3% of the rows had a different value than the norm since sampling may not reflect this -- however, is this the same thought process to use when all rows have the same value in a column?

Thanks!

	<Prev	Next>		<<First	<Prev	Next>	Last>>