On 04/09/2013 03:59 PM, Rafael Schloming wrote:
On Tue, Apr 9, 2013 at 2:00 PM, Michael Goulish mgoul...@redhat.com wrote:
I like your pattern -- { broadcast, balance, failover } ( addr, addr,
... addr )
idea a lot better than my multiple rules w/ same pattern.
I just think runtime-adaptive address translation could be very powerful.
i.e. for failover scenario, think of (old) Qpid clustering -- when one node
failed, a new node with a new addr stands up -- and advertises its addr.
It would get added to this rule
pattern -- failover ( addr1, addr2, addr_new )
And the dead node deleted.
I agree it is a very tantalizing possibility. I think application-specific
content based routing is the killer motivating scenario as that is
something where you couldn't use a turn-key router, and also not something
you are going to do via a config file.
Its my feeling that this functionality should be built on top of proton, not as
part of proton - exactly because its difficult to cover all the things that
people might want to do in a generic router. The routing library/plugin could
provide the kind of routing table discussed on the list by default, but could
also be presented as a toolkit that lets you plug your own routing code in, use
your own config files or database etc.