On 30/12/2011 20:42, Christopher Schultz wrote: > Ahmed, > > On 12/30/11 2:57 PM, S Ahmed wrote: >> I know with other frameworks (like python/rails) people tend to >> run multiple instaces of the web server and round robin requests to >> each using something like haproxy. > >> Is this known in the tomcat community at all? > > Are you asking if the Tomcat community knows how python/rails users > typically configure their servers? I would ask over there... > >> If I have a server with 16GB ram, would it make sense to run a few >> tomcat processes on different ports and use haproxy to round robin >> requests to each tomcat instance?
Is it a 64bit OS? Have you tested your application with a particular target number of users/requests in mind? > If your webapp is stable, I would run a single, large JVM. If it's not > stable, then running multiple JVMs will certainly increase your > redundancy. A multi-JVM setup also allows you to upgrade one webapp > instance and then the other to minimize (or eliminate) downtime -- see > your other thread on this subject. > > If you have multiple webapps, then the choice is up to you. In > production, we run separate webapps in separate JVMs -- that allows us > the most flexibility and protection against one webapp suffering some > problem like OOME and affecting the others. +1 Don't pick a model you like the sound of and try to squeeze your requirement into it. p >> I realize python/ruby do this because of their poor threading >> support. > > I have no idea. > --------------------------------------------------------------------- > To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org > For additional commands, e-mail: users-h...@tomcat.apache.org > -- [key:62590808]
signature.asc
Description: OpenPGP digital signature