Re: CLUSTER and clustered indices

Simon Riggs wrote:
> When a table has been CLUSTERed on a particular index AND that index
> values is monotonically increasing, then it would be a bad move to use
> blocks from the FSM since this would tend to destroy the natural
> clustering sequence.
>
> The index values will be monotonically increasing if a datatype is
> defined as SERIAL or if the default value is defined as the nextval of a
> sequence.
>
> Does anybody think it would be a good idea to not use the FSM if
> - we have a CLUSTER defined on an index
> - for the indexed column we have default value set of nextval()

That's a nice idea, but what's the cost? You will have to check every
insert: does the table has indexes? Is any of them clustered? Is the
clustered index attached to a sequence? It seems quite an expensive
check to be making.

--
Alvaro Herrera http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.