it will depend on your source data.
In some cases network bandwidth is an issue. If you have a wide interface and it is defined with fixed and it contains many variable length char target columns it can be a considered to change the source file as well to varchar as in this case the volume to be transfered can be drastically reduced.
Also consider if NO PI is an option in your approach.
Thanks a Lot Ulrich for the Valuable inputs !
I understood that If we can keep in sync the Source & target tables , the effective bytes would be reduced ( as in the case of varchar , the spaces would be trimmed off & in case of chars ,spaces got padded up and increase the length)..
Also in the file layout its better to define the field lengths as *VARCHAR inplace of *CHAR , as it will consider only the effective bytes rather than the max length of the column.
Please correct me if I am missing any thing here.
correct - at least if the AVG varchar length is < char length -2. As varchar will add up to additional 2 bytes for the actual length of the varchar (e.g. it need to know how long the value is to be able to separate different varchars)