[postgis-users] Is Index fetish bad?

[Ben Brehmer]

Hi All,

In what kind of scenarios should one not create an index on a table?

As I understand it, indexes can only be beneficial for SELECT 
statements. Unless of course the query planner inadvertently creates an 
inefficient query plan. I'm wondering how are INSERTs, DELETEs, and 
UPDATEs affected by the overhead incurred by index updates. Maybe the 
updating of the index is negligent in comparison to the actual INSERT, 
UPDATE or DELETE?

Is there any performance difference if an index is being updated for a 
thousand row table versus a 20 million row table?

Just wondering,

Ben