• @jeff, it sounds as if you should write a counterpoint article for EAV.

    Streaming stuff into an EAV has never been a problem, its the use cases for getting it out that are problematical.

    I can see it apply to telematics data to an extent in that I have an event type, date/time, value, GPS co-ordinate. Basically anything that is entered by machine reduces the data quality issues.

    One comment I didn't understand in the feedback in this thread was about SQL_Variant making rubbish of the points in my article. I know it wasn't you who said it but can you shine any light on this?

    Simple aggregations have worked fine, retrieval of all attributes for an entity have been fine. Pivoting I've seen eat a box.

    The particular example I gave where complex combinations of criteria caused no end of problems simply because the complexity of the SQL grew geometrically.

    One particular implementation was knocking on the door of 600GB so you can imagine the number of records in EAV form. It was a 9 figure number of records per day. We had to switch off mirroring due to the excessive IO involved.