I believe that its more efficient to group INSERT's together and COMMIT
them in groups.
However, I am automatically entering a lot of data and some of its is
not unique. How can I reap the benefits of using BEGIN and COMMIT
without the whole block of transactions failing if one is duplicated.
Than
Philip,
In order for psql to work out the most effective way to run queries it
relies on statistics about the size of your tables (amounst other
things).
Running VACUUM ANALYZE does two things:
The vacuum removes any tuples you have deleted from the database (before
this i think they are just mark
_query_object(param1
=> 'blah', param2 =>'foo')) as res(column_1 type_1, column_2 type_2,..)'
Then you execute this returned string. Problem here is two client-side
queries. Is there a way to do this all (code generation and execution)
server-side?
I am leaning towards solution 3 for now. But it's kind of a hack. Is there
any better solution that I am missing?
Thanks in advance.
Thanks,
Mat
asking for pointers about how to think about this problem
and any experiences people have had.
Thanks,
Mat
P.S. I am aware of limits listed here: http://www.postgresql.org/about/. I
am asking about practical size limits for performance consideration.
This will run on EC2 (or other cloud service) machines and on ssds.
Right now runs on m4.4xlarge with 64GiB of ram.
Willing to pay for beefy instances if it means better performance.
On Mon, Mar 28, 2016 at 4:49 PM, Rob Sargent wrote:
>
>
> On 03/28/2016 02:41 PM, Mat Arye wrote:
&g
