Thanks for your reply.
> > "When loading a table from a text file, use LOAD DATA
> INFILE. This is usually 20 times faster than using INSERT
> I wish they'd said whether their test used many single-row
> INSERT statements, or multi-statements. It could be
> that the latter approaches the speed of LOAD DATA INFILE.
> Since we can't trust that report, you should do your own
Yes true. It doesn't say compared to what, unfortunately.
> I think you're ignoring how slow hard disks are.
> Unless you have a fast RAID (as opposed to slow RAID, which
> still is pretty common) it's going to be at least a few
> times slower than your network link, so it *will* materially
> affect the upload speed. It's even worse if the MySQL
> server is on the same machine as the client.
I guess tempfs in /dev/shm would be a good candidate on Linux, as it is stored in-memory