We're told to take the load off the Database w/ Normalization: But how is Normalizing less "Load Intensive"?

Well let's take your pizza analogy a little bit further. What if you don't want to eat the whole pizza all at once? What if you want some tomorrow for lunch? Take it and put it in the refrigerator, right? This is sort of the same idea.

If you have a table with 4000 entries then when you do a search through that table it will look at all 4000 entries to get your results. If you can break that up into smaller tables with smaller amounts of information and link it back to a main point you can search your database in "bite-sized" chunks, so to speak :)

In essence, it makes searching and retrieving what you want more efficient. At least, this is my understanding of normalization.

Suppose you have a database of orders for a company. You could store the customer name, shipping address, and billing info separately for each order. But that is a lot of data to wade through in a search and store in the database, especially if a customer has many orders. Instead of redundantly repeating the same info every time, it's better to have separate tables for customers, addresses, and billing info and refer to them by their ids. It makes searching the order table easier because you aren't wading through redundant info. It make updating customer info easier too because you only need to update the data in the proper table once, and not go through every record in the order table to change every redundant entry.