Archives of the TeradataForum

Message Posted: Wed, 07 Oct 2009 @ 11:38:10 GMT

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

          > "(compressed columns allowed)
          >       fanned out into 26 hash join partitions,"

You can look up the phrases in the Transaction Processing manual.

The first phrase confirms that compressed columns can go into the spool file, which is just a good news.

"Hash join partitions" in this case have nothing to do with the partitions on the tables but are rather internal partitions on spool files created by Teradata to split the large hash table into 26 smaller chunks which can fit in the allocated memory. Optimizer takes these decision based on available resources and such DBC parameters as Hash Table Memory Allocation and Skew Allowance.

Number of hash partitions is interesting to monitor over the time. If you see significant increase, it could be a signal to tune the DBC parameters and / or add resources.

Victor

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