Re: Using Postgres to store high volume streams of sensor readings

On Fri, Nov 21, 2008 at 7:42 PM, Ciprian Dorin Craciun
<ciprian.craciun@gmail.com> wrote:
> On Fri, Nov 21, 2008 at 7:12 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> "Ciprian Dorin Craciun" <ciprian.craciun@gmail.com> writes:
>>> On Fri, Nov 21, 2008 at 6:06 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>>>> Not sure if it applies to your real use-case, but if you can try doing
>>>> the COPY from a local file instead of across the network link, it
>>>> might go faster.  Also, as already noted, drop the redundant index.
>>
>>>     It won't be that difficult to use a local file (now I'm using the
>>> same computer), but will it really make a difference?
>>
>> Yes.  I'm not sure how much, but there is nontrivial protocol overhead.
>>
>>                        regards, tom lane
>>
>
>    Ok, I have tried it, and no improvements... (There is also the
> drawback that I must run the inserts as the superuser...)
>
>    Ciprian Craciun.

    If I think better, the protocol overhead is not important...
Because if I don't use indexes, I obtain 600k inserts / second... (So
the test was useless... :) But I learn from my mistakes...)

    Ciprian.