[Sis-ams] another AMS spec tweak

Scott Burleigh Scott.Burleigh at jpl.nasa.gov
Fri Jul 25 17:47:02 EDT 2008


Tim Ray has suggested a small generalization of one step in node 
registration which would provide a bit more flexibility and perhaps 
scalability.  Because there could be over 250 nodes in a single cell, 
sending a single I_am_here MPDU containing MAMS state information for 
all nodes in the cell (as in 4.2.5.4.3.2) might conceivably result in a 
MAMS PDU that was too large for transmission given the MTU size of the 
underlying transport service.  It would really cost nothing in the spec 
to allow those node status structures to be clumped into groups of 10 or 
20 or whatever and sent in multiple I_am_here MPDUs that are much 
smaller, just as node status structures can be clumped into groups and 
sent in multiple node_status MPDUs as in 4.2.8.4.2.  The receivers of 
these MPDUs already have to be able to handle multiple I_am_here 
messages and also to handle multiple node status structures in any 
single MPDU.

So the proposed new text of 4.2.5.4.3.2 would be:

4.2.5.4.3.2    Optionally, the registrar may construct a set or one or 
more I_am_here MPDUs containing the MAMS state (i.e., MAMS endpoint 
name, all non-empty delivery vectors, and all current invitations and 
subscriptions) for all other nodes in the message space and send these 
MPDUs to the newly registered node.

No immediate effect on implementations since every implementation 
already does some proper subset of this, i.e., either sends one 
I_am_here or sends none (and sends I_am_starting MPDUs to all other 
nodes instead).

Anybody see any problems with this?

Scott




More information about the Sis-ams mailing list