How do you like your primary keys?

algorithm database-design relational-database primary-key ddl
Perpetualcoder picture Perpetualcoder · Dec 31, 2008 · Viewed 26.7k times · Source

In a fairly animated discussion in my team I was made to think what most people like as primary keys. We had the following groups-

  1. Int/ BigInt which autoincrement are good enough primary keys.
  2. There should be at least 3 columns that make up the primary key.
  3. Id, GUID and human readable row identifiers all should be treated differently.

What's the best approach for PKs? It would be awesome if you could justify your opinion. Is there a better approach that the above?

EDIT: Anyone has a simple sample/algorithm to generate human readable identifiers for rows that scales well?

Answer

Bramha Ghosh picture Bramha Ghosh · Dec 31, 2008

If you're going to be doing any syncing between databases with occasionally connected apps, then you should be using GUIDs for your primary keys. It is kind of a pain for debugging, so apart from that case I tend to stick to ints that autoincrement.

Autoincrement ints should be your default, and not using them should be justified.

Related questions

How do I check if an array includes a value in JavaScript?

What is the most concise and efficient way to find out if a JavaScript array contains a value? This is the only way I know to do it: function contains(a, obj) { for (var i = 0; i < a.length; i++) { …

Read More
Removing duplicates in the lists

Pretty much I need to write a program that checks if a list has any duplicates and if it does it removes them and returns a new list with the items that weren't duplicated/removed. This is what I have …

Read More
What does O(log n) mean exactly?

I am learning about Big O Notation running times and amortized times. I understand the notion of O(n) linear time, meaning that the size of the input affects the growth of the algorithm proportionally...and the same goes for, …

Read More