Question

If a sql server column is a string versus a guid, how would joins be impacted (assuming no indexes). Would it matter?

Also, when you put an index on a string column, does it become as efficient as an integer column with an index?

i.e. when you put an index on either a string or integer column, is the resulting index built the same way and therefore performs equally?

Answer 1

All other things being equal, less data is better. And by data I mean bytes.

For almost all SQL Server applications, the tightest bottleneck is disk I/O, and pulling less data from disk (or cache) makes everything faster.

This is variable depending on your declared string length. Bear in mind that GUIDs are 16 bytes, and varchar are 1 byte per character. nvarchar are 2 bytes per character. (n)varchar also have a 2 byte overhead per row to define the string length.

Space/bytes wise, a String is bigger than a GUID is bigger than an int.

The smaller/tighter your field definition the better, so int is faster than a guid, which is faster than a string.