Feb
28
Expensive database server = 1980s Filofax
The way databases actually store data on their hard disks is pretty important in high performance situations. It's typically thought of as arcane, and few people understand it.
You can think of a database as being like a 1980s Filofax address book. You fill the pages out in ink - changes are crossed out and re-added.
You can fit all your contacts in a few pages, or you can spread them out, one contact to each page. This is called fill factor.
If you pack them together you only need to check a few pages to find the contact you want. However when adding contacts you have to add two new pages, one with 1/2 the contacts, and another with the other 1/2 + your new one. When changing a contact you have to cross out a contact, but then you have to add a new page and copy it all out again.
You can think of a database as being like a 1980s Filofax address book. You fill the pages out in ink - changes are crossed out and re-added.
You can fit all your contacts in a few pages, or you can spread them out, one contact to each page. This is called fill factor.
If you pack them together you only need to check a few pages to find the contact you want. However when adding contacts you have to add two new pages, one with 1/2 the contacts, and another with the other 1/2 + your new one. When changing a contact you have to cross out a contact, but then you have to add a new page and copy it all out again.
