Archives of the TeradataForum

Message Posted: Wed, 08 Aug 2007 @ 20:17:20 GMT

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

1: Personally I like mload over fastload (I use both). I think the error checking is a lot better in mload and easier to find errors in the table. I know other's prefer fastload.

2: Regardless of whether I use mload or fastload, I use an interim staging database and table to load the data into the system. Once everything is loaded successfully into the staging area, I programmatically move the data into my production table. Using this method also, I can use a multiset staging table which expedites loading and then use an SQL to load to production doing any validation, duplicate checking, etc.

3: Over the years, I have come to prefer to add a date column just for my own purposes to any data table which is appended to on a routine basis (as opposed to dropped/loaded). This way I can very quickly look for missing load days (programmatically) and also quickly back out data if need be for a single day. Since I load at night for all of our jobs, I make anything loaded in the evening prior to midnight the next day's date and anything after midnight the current day's date. This addresses the challenges of receiving data some days at 12:01 AM and the next day at 11:59 PM.

I do a lot of other error checking and routine validation, but I think this covers your questions. :-)

These are just some of my thoughts/preferences.

Craig

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