On Tue, 21 Oct 2003 10:15:45 +0800 (CST) "Zhu, Yi" <yi.zhu@intel.com> wrote: > Reliable multicast transport layer is useful in the intranet. > If kernel can provide this, a lot of cluster softwares can > benefit from it. Who cares? Nobody uses ipv4 multicast. Do you know how I can tell? The fact that I _NEVER_ see bug reports for ipv4 multicast usage yet I know there must be bugs lurking there. The only time people actually use any kind of multicast is via the automatic network scanning mechanisms of IPV6, and typically these show up ethernet driver multicast bugs. These ethernet driver bugs in multicast handling are usually so severe that it is clear that it would be impossible for ipv4 multicast to work at all on such cards. Therefore, it is clear that at a minimum the set of people having such cards were not trying to use ipv4 multicast. Multicast on ipv4 had a lot of potential, but (as stated by those wiser than I, such as Jacobson) the web "killed the internet" and that's why we're left with what we have today as a content distribution mechanism (HTTP) instead of something more distributed and intelligent (DNS). People on clusters use their own special clustering hardware and protocol stacks _ANYWAYS_ because ipv4 is too general to serve their performance needs. And I think that is a good thing rather than a bad thing. People should use specialized solutions if that is the best way to attack their problem. - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
