I had the same initial reaction in reading the post. My assumption was that indexing would have sufficiently sped up the query speed problem. However, normali

I had the same initial reaction in reading the post. My assumption was that indexing would have sufficiently sped up the query speed problem. However, normalizing the fields that contain repetitive data into separate tables could create significant space savings since the full text for each column would not need to stored for each row. Instead of 20-30 bytes per email address, a 4 byte (assuming 32-bit era) OID is stored in its place. It's pretty easy to imagine how quickly the savings would add up, and that it would be very helpful in the era before SSDs or even 1TB HDDs existed.