Archives of the TeradataForum

Message Posted: Wed, 05 Nov 2003 @ 09:15:46 GMT

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

Victor,

If such data only forms a "...small part of the data in the table..." is using Unicode such a big overhead ?

Having said that, have you tried/considered storing such data ("text_info" in your example) using the BYTE data type?

There are obvious limitations when using BYTE columns, but it also has some advantages compared to Unicode. As I see it these are:

Pro's

- there are no cpu cycles spent converting data when you load it to Teradata or read it back from Teradata

- disk space rqts are approximately half that of using Unicode (but as noted above is that a major problem in the overall scheme of things)

Cons

- you can't join this column to any other columns unless they are also BYTE/VARBYTE,

- you can't CAST this column to any other data type,

- you won't be able to read the data using 'simple' tools such as Queryman or Bteq. As you note it will be down to the client s/w to interpret the data correctly.

- if you're using an ODBC application to read the data I don't know if the odbc driver will do anything funny with the data (I'm guessing that it won't, but I simply don't know).

I've only used BYTE in one situation (in @10 years), in that case I simply needed to get a key value into and out of Teradata and use it match up with key values back on my host. There were probably other ways of doing the same thing, but we used BYTE to achieve what we wanted.

Regards,

Dave

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