[jboss-cluster-dev] TCP vs. UDP

M Wagad wagad at hotmail.com
Wed May 14 11:36:24 EDT 2008


Thanks Brian. Will do

> Date: Wed, 14 May 2008 09:53:18 -0500> From: brian.stansberry at redhat.com> To: wagad at hotmail.com> CC: jboss-cluster-dev at lists.jboss.org> Subject: Re: [jboss-cluster-dev] TCP vs. UDP> > Please post on the JBoss Clustering User Forum at > http://www.jboss.com/index.html?module=bb&op=viewforum&f=64 . This > mailing list is for discussions related to the development of clustering > features in JBoss projects, not for usage questions.> > M Wagad wrote:> > We have 2 node JBOSS Cluster, and we are having performance issue, what > > we found out is that if we turn of the slave server, and operate the one > > master node (Changing the load balance to 100/0), we have much better > > performance then when we have both of the running sharing the load with > > 50/50 load balance.> > > > Would changing the default Protocol of UDP to TCP make any difference in > > terms of Performance?> > > > I had come across these posts on the Net that praises TCP over UDP for > > organizations that have less then 16 JBOSS Nodes :-> > > > > > > > > > For TCP…> > > > > > > > /“We experiment with another configuration using TCP/IP and show that > > current J2EE application server clusters up to 16 nodes (the largest > > configuration we tested) can scale much better with this configuration. /> > > > /We attribute the superiority of TCP/IP based group communications over > > UDP/IP multicast to a better flow control management and a better usage > > of the network switches available in cluster environments.”/> > > > > > > > And UDP..> > > > > > > > “Having to support reliable multicast UDP communication is not a piece > > of cake. Drop a couple of packets and you get only a fraction of the > > throughput. Then lose the ordering of the packets and you are in a > > slump. The situation can turn pretty sour if you use large messages that > > get fragmented (more than one UDP packet). You only need to lose one > > packet to lose your entire message - then you resend everything. The > > point is that in unideal and congested conditions you start to miss TCP, > > and it is probably extremely hard to beat its adaptive protocol that has > > been perfected for so many years.”> > > > > > > > > > > > I would also like to know if any companies out there make the change > > from UDP to TCP and if they can share their experience with us.> > > > > > > > Thanks> > M. Wagad.> > > > > > ------------------------------------------------------------------------> > > > > > ------------------------------------------------------------------------> > > > _______________________________________________> > jboss-cluster-dev mailing list> > jboss-cluster-dev at lists.jboss.org> > https://lists.jboss.org/mailman/listinfo/jboss-cluster-dev> > -- > Brian Stansberry> Lead, AS Clustering> JBoss, a division of Red Hat> brian.stansberry at redhat.com
_________________________________________________________________

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/jboss-cluster-dev/attachments/20080514/060b3765/attachment.html 


More information about the jboss-cluster-dev mailing list