On 05/28/2010 02:55 PM, Craig Ringer wrote:
> On 28/05/10 17:41, Joachim Worringen wrote:
>> Greetings,
>>
>> my Python application (http://perfbase.tigris.org) repeatedly needs to
>> insert lots of data into an exsting, non-empty, potentially large table.
>> Currently, the bottleneck is with the Python application, so I intend to
>> multi-thread it.
>
> That may not be a great idea. For why, search for "Global Interpreter
> Lock" (GIL).
>
> It might help if Python's mostly blocked on network I/O, as the GIL is
> released when Python blocks on the network, but still, your results may
> not be great.
I verified that the thread actually execute queries concurrently. That
does imply that they are blocked on I/O while the query is running, and
that the query performance does in fact scale for this reason.
In the "import data" case, however, I really need concurrent processing
on the CPU in the first place, so you may be right on this one. I'll
check it.
>> will I get a speedup? Or will table-locking serialize things on the
>> server side?
>
> Concurrent inserts work *great* with PostgreSQL, it's Python I'd be
> worried about.
That's the part of answer I wanted to hear.,,
thanks, Joachim