This depends on the number of indices. Insertions into a B-Tree
essntially follow an Nlog(N) law, so the best you can expect is
O(Nlog(N)). Similarly for deletions.
A SELECT depends upon the query and indices. A row search is
essentially linear with size, O(N), whereas a B-Tree index search is
enormously faster, but does follow a basic O(Nlog(N)) law.
The conclusion is that small N is always faster in a tree or list
structure. A table for each data type may be preferable to all types
being lumped into one table and selected on type.
JS
Hannes Ricklefs wrote:
Hello,
does anyone have any experience in the amount of time increase for an UPDATE,
INSERT, SELECT in relation to the size of the actual database?
Is it proportional or exponential for example...
Regards,
Hannes
