Archives of the TeradataForum

Message Posted: Thu, 15 Feb 2007 @ 13:45:21 GMT

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

<-- Anonymously Posted: Thursday, February 15, 2007 08:47 -->

Your problem is that Teradata does a "tag sort", appending (another copy of) the sort key to the spool rows; even worse, the "tag" fields are fixed length (converts VARCHAR to CHAR).

You could probably SUBSTRING 16K or so of the text. Or use a UDF to generate a hash or "message digest" that would approximate uniqueness, GROUP BY that and use MIN or MAX to aggregate the long text.

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