> The problem with performance sheets is that there are so many
> influencing factors that the information is all but useless. If you
> want a ballpark figure, you should be exceeding 10,000 inserts per
> second for a prepared query inserting records into a reasonably simple
> table.
>
> But add several indices to the mix, add some constraints, add a bunch
> of insert triggers to the mix, force your database to compete for CPU
> or disk access, chuck a slow network between your client and database
> server, make some of the fields really large varchars, hold your
> transactions open for so long that garbage builds up etc can quickly
> tumble performance.
>

I agree there´s many influencing factors. But, some info to guide new
users should help. Can this be done using a DB example, in a local
machine with a predefined configuration. Example: "in our tests, in a
Pention D 3.2 GHz, with 1 GB DDR3 RAM, we reach averages of: xxx
inserts per sencond in table xxx not prepared, xxx inserts in table
xxx prepared, xxx deletes..., xxx inserts and deletes ...... with the
Example DB."

> Without providing many clues about your particular environment, I
> can't know whether your experience is expected or indicative of some
> other problem. I can assure you there are people on this list that
> would consider your 30 million record "huge" table relatively small.
> Our largest table (record number wise) is heading towards 10 million,
> and inserts are measured in milliseconds (even with a trigger and two
> foreign keys to check), so if you are wondering about "minutes" for
> hundreds of thousands of records, then there must be other causes.
>
> Can you provide your database engine version as well as the output
> from gstat -h?
>
>
> Adam
>

Like I said, mine db is small and fast. But I know it will grow up.
I´ll try to get the infos from the db with 50 millions records.
thanks for reply.