If database insert speed is the bottleneck, and you have clean data, one of the best ways to gain speed is to drop indexes and constraints on the table before you insert, then re-create them afterwards.
This presumes, of course, that you have permission to execute DDL in the database, that you know the table schema, that you won't hose up your data by not having constraints, and so on.
Also, some databases, like Sybase and SQL Server, have specialized bulk-insert interfaces that will allow you to load data irrespective of constraints. But that may not work for you, since you seem to indicate you're going against multiple platforms at different times.
In reply to Re: Implementing a buffered read-and-insert algorithm
by VSarkiss
in thread Implementing a buffered read-and-insert algorithm
by radiantmatrix
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |