[atoca] what needs to be done yet.

["Mark" <[e-mail]>]

Hi Guys,

The features that make an emergency Public Warning Messaging special are; 

1 the super-massive scale of the broadcast 
(Maybe to 10s of millions of hosts)
 
2 in a handful of seconds 
(CMAS takes only 7 seconds to deliver them all),
 
3 during extreme overload, 
(CMAS uses out of band signalling)
 
4 without any foreknowledge of their addresses, 
(No data base needed by CMAS)
 
5 and in a completely stateless passive way, 
(No TCP session exhaustion possible)
 
6 which is geo specific, 
(CMAS is selective down to one cell, about 1 mile)
 
7 Must be highly resistant to Slashdot or DOS events. 
(Disasters are always without exception accompanies by Tsunami waves of
load). 

Very extensive studies carried out by both government and industry showed
that the most appropriate way of dealing with these very extreme events is
to use passive point-to-multipoint bearer services. For example in the
cellular networks, Cell Broadcast and multi media Broadcast multicast are
such bearers. This is why they had to be deployed for this special
situation. The studies were very thorough, and completely conclusive.

The closest equivalent to cell broadcast in IP is IPMulticast.  

These situations are not at all the same as the situation where CAP messages
are being signalled between relatively small numbers of well known systems
which are in a trusted relationship with each other, and there is 'stateful'
handshake relationship between hosts. This is why IMHO a completely
different philosophy is needed, and one which I don't think ATOCA have
looked at in depth yet. 

However the title of the WG clearly indicates that we should, and so as Art
says, the reality on the ground has only just this year caught up with the
subject. Well we WERE ahead of the subject, but not now!

IMHO there is work to do and we have not done it yet :-)

Mark Wood DRCF.