[Standards] Council on Stanza Repeaters without Multicast
Carlo v. Loesch
CvL at mail.symlynX.com
Fri Apr 4 13:43:37 UTC 2008
| > A quick data-point: the largest pubsub node *right now*, has a bit
| > over 73000 (73056 to be exact) subscribers, the top 3 are all above
| > 72k. And this service is only for local users of the server.
Dave Cridland typeth:
| To state the obvious (which I'm sure you know), nothing's going to
| help this case...
| ... unless you ran a repeater over the component link, of course,
| which might work. Or might not - but a component protocol might need
| this sort of thing.
Now this is a good example of what I am critizing. By making specific
multicast tree extensions to each application of XMPP you always run
into new applications that still need such a fix.
If you have a generic routing level solution for smartcast/multicast
distribution instead, and slowly optimize your applications to it,
then any one-to-many situation can be solved with it, like this one.
If I was strategic about PSYC, I should recommend your strategy of
messing up each part of XMPP piece by piece instead of cleaning it up.
That would guarantee, that even in ten years PSYC is perceived as a
much cleaner and organic approach.
It's your choice.
More information about the Standards