Search results for query: *

Does doing a fastexport/multiload into an empty table perform faster than an insert/select into an empty table?

Here's what I get when I right click on the table and select indexes: IndexName ColumnName Type Unique IndexNumber ColumnPosition OID Primary Y 1 1 FundCode Join Index N 8 1 Appropriation Join Index N 8 2

Here are the results - we have tried doing a drop index on index 8 but it won't let us: IndexNumber IndexType UniqueFlag ColumnName ColumnPosition 8 J N Appropriation 2 8 J N FundCode 1 1 P Y OID 1

I can only see it when I right click on the table and show indexes. When I show the table ddl, it doesn't show up. I have no idea how it got there either.

In trying to clean up a test database, I've discovered a join index with no name on one of the tables. Does anyone know how to drop this index? I've tried drop join index but since it doesn't have a name, I'm stumped. Thanks for any help.

You were right in that the original statistics on the tables were collected before there was any data in them. I made my own copies of the tables etc. I couldn't even create the index with the order by hash clause because teradata recognized that there already was an index on that column...

From reading the manuals, I gather that order by hash and order by values clauses on indexes help when querying a range of values. Or am I totally off base here? So I have a table with a UPI - one column numeric. A colleague of mine added a NUSI on the same column with the order by hash...

You are right again - thanks.

In the following stored procedure I have the table t_lookup_changes_temp defined as a perm table and the procedure works. When I try to define the table as global temporary, the procedure runs but does nothing. Any ideas? I'd like to use global temp because I may have several procedures...

Why, when you add referential integrity to a table and there are no errors, doesn't the system table get dropped. It's confusing - it always makes me think there are errors. Just curious.

Thanks - that worked.

Thanks - seeing the errors helped somewhat. But the compiler doesn't like the declare cursor statement and I don't know why.

I'm trying to compile a stored proc which contains a cursor. When I compile the procedure in queryman, it doesn't give me error messages. Here the procedure - any help would be appreciated: Replace PROCEDURE ncr_temp.spCpaisUpdate( ) BEGIN --...

Thanks Dieter - I had thought about the insert/select but space is at a premium. I like the export/multiload idea better - some of the columns being updated have secondary indexes on them but the primary key is not involved at all. Here's what I'm trying to do. There are a bunch of lookup...

I need to update rows in a large (100 million rows) table. The number of rows that will be changed is relatively small compared to the size of the table - my guess is that no more than 2-3% will be updated at any one time. Does anyone know of a fast way to do this other than with update statements?