On Fri, 2009-05-08 at 09:36 -0700, Robert Hodges wrote: > So far in this thread nobody has brought up the issue of shared locks, > intent locks, and lock promotion. These are three further bits of > nastiness > that make deadlocks both more likely on the slave and also almost > impossible > to reason about in advance. IMHO any attempt to achieve > parallelization in > the general case is therefore doomed to failure.
Yeah its not really looking good for the original idea is it? ;) While certainly too coarse for some apps, am I crazy in thinking that the LOCK TABLES method would provide a simple "free" boost for multicore slaves that do need to operate on multiple tables? Anyway.. the only other gasp for air that I can reach for to get beyond that single table thread limitation is to rely on row based replication, and serialize on table+rowid But I haven't given that much thought, so I'll just have to think about it more... _______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
