Re: Tuning the import buffer parameter

You are effectively increasing the amount of data that can be loaded per commit, which is why you see better performance with larger buffer sizes.

However, make sure you have sufficient *free, real memory* and that no other applications need some of that memory before you start running very large buffer sizes, else the whole lot will start paging.

There is also a cut-off in terms of performance improvements you'll see with progressively larger buffer sizes. Test your loads to find the 'sweet-spot' for your system.

Steve Phelan.

Matt Larson wrote in message <6e4ha8$epv$1_at_server1.incc.net>...
>Does it matter how large I set the buffer parameter on an import (from a
>speed perspective)? I have one table in my database with a 5meg row. All
>other tables can be imported with buffer=1000000. Should I import the rest
>of the tables with the small buffer, or just raise the buffer size for the
>whole import. What does this parameter really do? Would it hurt to set it
>to 10M, 100M, 500M??
>
>Please email answers to
>matt_larson_at_jdedwards.com
>
>Thanks,
>Matt
>
>
Received on Wed Mar 11 1998 - 00:00:00 CST