Archives of the TeradataForum
Message Posted: Tue, 13 Jun 2000 @ 17:37:00 GMT
We have a WareHouse in Teradata V2R3 with 10 nodes, 104 Amps and 20 PE, memory = 14 Gb .
We have a table with 9.000.000 of rows, the primary index is a UPI, the distribution is good on all amps, but there is a column (date), thats accept nulls. We have three kind of values in this column: nulls (4.100.000 rows), 99/01/24 (3.900.000 rows) and another dates.
There're very complex queries with a condition "column is null" that involve that column. When we run two or more queries like this, the Teradata performance is degraded, because a hot amp is caused by a wrong distribution in spool space.
We analyzed the model and we can't modify it. This table is one of the most important in our data model, we can't change it and that condition is useful for the final users. We thought an alternative query with a condition "not (column is not null)" but we still have a hot amp because there're 3.900.000 of rows with the same value(99/01/24). This problem appeared after the upgrade of V2R2 to V2R3.
Is there any way this query can be run without degrade the performance? Why this Hot amps are produced?
|Copyright 2016 - All Rights Reserved|
|Last Modified: 28 Jun 2020|