On Wed, Oct 03, 2001 at 06:46:52PM -0700, Nathan Meyers wrote: > On Wed, Oct 03, 2001 at 06:32:11PM -0700, Dan Kegel wrote: > > root wrote: > > The 2.4 kernel uses 32 bit process ids, so that shouldn't be a > > problem. Are there other precious resources you're worried about? > > If not, there's a good chance NxM threading models will simply > > introduce complexity rather than add performance. That was Linus' > > bet long ago, and the question is still open. > > I'll defer to anyone who's faced this problem head-on. Any opinions > from those who've hit the wall when scaling up Java on Linux?
NxM is actually pretty well proven as being a better way to scale. It's also been proven that it's fairly complex to implement properly. In general, there are a lot of issues with having N kernel-level threads running where N is a large number. Generally speaking, it works great as long as N is roughly proportional to the number of processors in the system. Unfortunately, until JDK 1.4 came out, Java programs would have threads proportional to the number of simultaneous I/O operations (which in a network application is typically MUCH higher than the number of threads). As far as hitting the wall on Linux, without a lot of tweaking, you do run into problems with memory consumption once you start spawning a lot of threads. --Chris ---------------------------------------------------------------------- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
