> On Oct 10, 2018, at 1:24 PM, Andres Freund <andres(at)anarazel(dot)de> wrote:
>
> On 2018-10-10 18:50:02 +0200, Laurenz Albe wrote:
>> Rob Sargent wrote:
>>> Can anyone here tell me whether or not the CopyManager facility in JDBC
>>> via org.postgresql:postgresql:42.1.4 is internally multithreaded?
>>> Running on CentOS 7 (all participants), java8, postgres 10.5
>>
>> It isn't, and there would not be much reason for it to be, as COPY
>> in PostgreSQL cannot be parallelized.
>
> s/cannot/is not/.
>
> You obviously can just copy the data into postgres over multiple
> connections if you need to speed COPY up. But that requires splitting up
> the data on the clientside.
>
> Greetings,
>
> Andres Freund
Interesting note, deeply pocketed. The COPY portion is /not/ my problem - that’s plenty fast enough and memory efficient enough straight from the box. I had over looked that I was requesting 33M records, in a map with a longish CSV string as key...then playing sort games with keys and ... well that’s as stupid as I got.