Re: Database character fields

Mike

I'm answering late - and Chuck probably gave you the truth!

Anyhow, variable length fields on iSeries are implemented using an auxiliary space - the fixed portion is in one MI space object, the variable data is in another. The fixed part has "pointers" of a sort - forget if they are real pointers or offsets or what. They refer to places in the auxiliary space or spaces.

The result is, you can have 2 physical IOs for each varying length field - not just record.

OTOH, since data is read into memory in 4K pages, the data might already be in main memory when it is needed, hence, not needing the extra physical IOs.

Now I don't know how MySql does it - there are other ways of doing it, I have to think.

HTH
Vern
-------------- Original message --------------
From: Mike Cunningham <mcunning@xxxxxxx>

Not looking to start a war over which is better, just looking to update my
knowledge.

When we create character fields in DB2/400 we usually setup fixed length, no
null support fields. If the character fields is long (e.g. 100+ bytes) and we
suspect the data will vary greatly in actually used length, we will make it a
variable length field. The only time we use null fields is for data we are
importing in from other systems where the file can have null values. In those
cases the file with null field support is usually a work file used for the
import and they we move the data into non-null fields in the production files.
In other databases (like MS SQL) the standard looks like it is just the
opposite. All character fields are variable length with null support unless you
take extra steps to not do that. My training (and it has been some years) said
that variable length fields are good for saving storage space but bad for
overhead. That the database had to do extra work to manage the variable length,
tracking the actually number of bytes in use and manage the ov!
erflow areas when the data in the fields changed from 10 characters to 1,000
and back to 10. Is it still true that variable length fields are less efficient
and if so why do other databases have that as the default? Or is this something
specific to the implementation of the database? Is DB2/400 move efficient with
fixed length but MS SQL more efficient with variable length?
--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
To post a message email: MIDRANGE-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/midrange-l
or email: MIDRANGE-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/midrange-l.