Dan Colish wrote:
> CREATE TABLE
> INSERT 0 100000
> Timing is on.
> COPY 100000
> Time: 83.273 ms
> BEGIN
> Time: 0.412 ms
> TRUNCATE TABLE
> Time: 0.357 ms
> COPY 100000
> Time: 140.911 ms
> COMMIT
> Time: 4.909 ms
>
>
>
Anything that doesn't have times that are orders of magnitude greater
than this is pretty much useless as a measurement of COPY performance,
IMNSHO.
In this particular test, to check for paring times, I'd be inclined to
do copy repeatedly (i.e. probably quite a few thousand times) from an
empty file to test the speed. Something like:
select current_timestamp; begin; truncate; copy;copy;copy; ... commit; select current_timestamp;
(tests like this are really a good case for DO ' something'; - we could
put a loop in the DO.)
cheers
andrew