On Friday 15 Feb 2002 7:24 pm, David Hamilton wrote: > Nicolai P Guba wrote : > >> On Friday 15 Feb 2002 2:47 pm, David Ward wrote: > >> > >> 2) Take a look at the Reflection Performance improvements graph here: > >> http://java.sun.com/j2se/1.4/performance.guide.html > >> Again not sure, but this looks like it could really help JBoss > >> performance with the Proxy layer to the EJBs. > > > > Probably not very much, unless deploy-time becomes of an issue. I cannot > > imagine any run-time features that would require relfection. Or? > > Think it depends what you mean by reflection. If you are referring to the > process of getting Method and constructor objects the I agree, but this > seems to refer to the actual invoke method. > > This article that the invoke method is 20 (twenty) times faster in 1.4 than > 1.3.1. > > There must be at least two invocations on any EJB call (Proxy and > Container), and that would, in the past, have added a much larger overhead > than a straight method call.
Indeed, this is good news. > Looking at the rest of the article, the claims it makes are quite > impressive (and if realistic, would make a significant difference to JBoss > performance, IMHO). > They are (over 1.3.1): > Servlets: +35% > EJBs: +34% > SPECjbb2000: +58% > > Anyone fancy trying some figures for JBoss? I'd agree on the impressiveness ;) We are going to stress test our servers, if there is any useful data, I'll publish them here. cheers! _______________________________________________ JBoss-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/jboss-user
