Re: Normalization, Natural Keys, Surrogate Keys

On 21 May 2002, ed.prochak_at_alltel.com wrote:
> Galen Boyer <galendboyer_at_yahoo.com> wrote in message

>> Performance should be increased on joins with tables each having one
>> surrogate key over tables all having natural keys and most of these
>> having multiple values.

[...]

> (now there is a factor of the size of the keys, so a table with MANY
> rows will be slower for the natural key because fewer of them fit in a
> block, so more blocks must be read to find it. but if this amounted to
> something even near a factor of two, I'd be supprised.)

This is what I meant. Scan performance are all about the amount of blocks being read, index or table.

-- 
Galen deForest Boyer
Sweet dreams and flying machines in pieces on the ground.