On Wednesday, April 22, 2015 at 2:46:19 PM UTC+1, Michael Bayer wrote:
>
>
>
> I carefully re-read your first email, and now I see, that you are trying 
> to write *TO* the database, yes?    And you're comparing the speed of 
> writing *TO* to the speed of reading *FROM*? 
>
> Unfortunately, selecting data from some tables compared to inserting 
> data in those tables are two completely different activities.   It is 
> not at all surprising to see a large difference in speed. 


Of course. I do not expect that reading from will be as fast as writing to 
the database.
However, 6 minutes to write 11MBs of data is ridiculous, because this is 
the time it takes even when:
1) there are no text columns (so the nvarchjar(max) issue I mentioned and I 
saw in other cases cannot apply)
2) there are only floating-point numbers
3) there are no primary keys, no indices and no constraints of any kind

I trust you'll concur it's hard to justify such a long time to transfer 
such a small table even in the examples above.
I am not a DBA, but I cannot think of other reasons that would cause such 
terrible speeds. Any thoughts would be more than welcome!

Maybe the to_sql method is committing to the database after inserting every 
single row and this slows things down? Even so, 6 minutes for 11MBs means 
31KB per second. My internet connection in the '90s was faster than that :)

I have also been very surprised by the lack of info on the net. Either I am 
the only one using the to_sql method, or I am the only one who finds it so 
slow!

>
>

-- 
You received this message because you are subscribed to the Google Groups 
"sqlalchemy" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sqlalchemy.
For more options, visit https://groups.google.com/d/optout.

Reply via email to