If in super fast mode, it will show "(qvd optimized)" next to the load. I also thought it would show it in the log, but I'm not seeing it, which seems strange.
Expanding on this, you can rename fields, and you can have a single where exists() as your condition. I believe that anything else will break the optimized load, but I'm having a hard time finding confirmation in the help text.
Since you can only have a single where exists, you'd want to use it on the most restrictive field to get the fewest rows back. Other conditions can be applied after the fact with an inner join, and this is often faster than breaking the optimization on the original load. If performance is critical, try multiple approaches and keep the fastest, of course. But where exists followed by an inner join is my standard these days. Often you might need to do a little prep work for the where exists. Like if you only want YTD data from a file with 10 years of data, load a calendar with YTD dates, and then use where exists(Date). That sort of thing.
The QVD itself is not 'optimized'. It's a matter of how you do the load, now how you make the QVD. Though as hopkinsc indicates, if you put necessary transformations in the QVD itself, you won't have to do them when loading from the QVD, so this will support using an optimized load. In that sense, I supposed you could say that that QVD has been optimized.
