Archives of the TeradataForum
Message Posted: Fri, 16 Feb 2001 @ 19:53:57 GMT
| Subj: || || Re: Collect Stat |
| From: || || Steve Bracewell |
What really works well is automating this entire process. We run an automated process that uses a 'cycle' table to determine which of
our non-fastloaded tables are due for collected stats (as determined by demographic change frequency and 'last date collected'). The
process then queries DBC tables to produce collect stats statements for 'stale' single and multi column nupi and upi indexes. The 'cycle'
table is managed by dbc.tables, alleviating the possibility of a table being introduced without making it to our 'cycle' table. This, along
with our other administrative processes we've automated, leaves us time to do really cool things.