On Mon, 2006-04-17 at 14:22 -0500, David Nicol wrote: > Maybe someone could write Apache::DBI::pooled which would > maintain a smaller number of persistent connections and provide > them to the client processes as needed intstead of maintaining > a persistent connection for each client process.
It's possible, but it would be difficult. Sharing database handles between processes would depend on how the client libraries work, so I expect it would need to be implemented separately for every database. Stas did some work on making DBI handles shareable across threads, but you'd still need each DBD to add support for it, I think. If you're interested in pursuing it, the DBI mailing list can probably offer more informed opinions about it. - Perrin
