Some commenters have asked what kind of business would deploy this kind of technique.
I’ve done it. For MLS syncing software, there are lots of properties, not thousands but can be hundreds. And each MLS RETS has its own Schema, so for any kind of logic portability this is necessary.
Actually, storing the raw data as a blob is a flexibility technique and is a separate concern than the number of fields. As I can’t predict the future set of optimized queries I’ll need, and I don’t want to constantly sync and resync (some MLS will rate limit you), then this way I can store the raw data once, and parse plus update my tables/indexes very quickly.
I’ve done it. For MLS syncing software, there are lots of properties, not thousands but can be hundreds. And each MLS RETS has its own Schema, so for any kind of logic portability this is necessary.
Actually, storing the raw data as a blob is a flexibility technique and is a separate concern than the number of fields. As I can’t predict the future set of optimized queries I’ll need, and I don’t want to constantly sync and resync (some MLS will rate limit you), then this way I can store the raw data once, and parse plus update my tables/indexes very quickly.