Archives of the TeradataForum

Message Posted: Wed, 05 Jul 2006 @ 17:42:31 GMT

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

1) Secondary indexes are logically spread across all AMPs. (In the extreme, skew could cause some AMPs not to have SI data.) All AMPs will use the same "subtable ID" for blocks of the same index.

2) For USI, distribution of the entries is based on RowHash of the SI value - so the USI entry will generally be on a different AMP from the base row to which it points. (For NUSI, index entries will always be on the same AMP as the rows to which they point.)

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