We are trying to snapshot data from 8.1.7.4 database with single byte
8bit characterset (WE8MSWIN1252) to multipbyte (UTF8) also running
8.1.7.4

However we have realised that any field sizes are multipled by 3 to
accomodate any characters that will require 3 bytes to be stored in
their equivalent form on a multi-byte character set database e.g. Euro
character

This creates a problem with large keys because they are hitting the
limit of 3218 bytes available for the PK in a 8k block size.

Is there any way of preventing this multiplication of the field sizes
?