What is the best way to deal with AutoNumber function on QVD files?

[mwallman]

Hello all,

What is the best way to deal with autonumber function when it comes to QVD files?

We are currently loading QVD files, and then using AutoNumber function on the key fields. I think this makes the QVD files non-optimised. What is the way around this issue?

Our general process is to load data from source > convert it to QVD files > transform it > load into a QVF > binary load the QVF to a Dashboard app.

Suggestions welcome.

[dwforest]

If you have a key field, why AutoNumber?
If you are combining fields to make the key, i use Hash128 function to generate a unique key per combination and use resulting key as keyfield.

[marcus_sommer]

Autonumber() is a very useful feature but it shouldn't be applied for each KEY respectively it will have side-effects which could reverse the benefits into disadvantages.

Depending on your data it might be an option to create the autonumber already by the creation of the qvd's - maybe by pre-loading a distinct fieldvalue-list from the other data-sources to which the later qvd should be associated - but this needs some more logic and working-steps as an usual etl-approach.

If you couldn't apply such a workaround you need to decide if the autonumber-key has more benefits within an application as the loss of the optimized qvd-load.

- Marcus