Splitting the Table

Normalization's redundancy savings are countable — splitting a combined table into two linked by a foreign key replaces repeated copies with single records, and one small case shows the arithmetic.

Start with one orders table holding six orders from three customers, each row carrying the customer's full address. The address data appears six times — once per order — and a customer moving means finding and fixing every copy, with any missed row breaking consistency, video 0416's third test. Now normalize. Split into a customers table and an orders table: three customer rows, each holding the address once, with a chosen primary key (0421); six order rows, each carrying only the customer's key as a foreign key (0422). Count the address copies: six before, three after — halved in this tiny case, and the ratio improves as customers order more. Count the update work: a move now touches exactly one row, and consistency holds by construction rather than by diligence. The orders lost nothing — the foreign key threads each one back to its customer whole. Redundancy down, integrity up, which is normalization's definition (0427) executed in six rows.