[Sis-csi] RE: networking details

[Keith Hogie]

Adrian J. Hooke wrote:

> But isn't that in fact the reality of the budget situation?  Even within 
> the most rosy scenario (below), is it really clear to everyone that we 
> need a fully-routed infrastructure prior to about 2018? So shouldn't we 
> be planning an evolutionary strategy that migrates us gracefully from 
> today's systems towards some future Nirvana? More to the point, if we 
> don't need routing for another decade, do we need to pin down all of the 
> protocol details *now*?
> 
> 
   Moving to spacecraft using Internet protocols a change to the whole
space communication concept.  We have been using Internet technologies
in our ground systems (control centers and data processing facilities)
for 10 to 15 years.  Our data all moves around the ground using Internet
technologies.  Now we are changing the space end so that selected
Internet technologies and be used end-to-end.

If we accept that we want and need a routed infrastructure in space in
the future, why wouldn't we want to start putting it in place with
missions currently being built. If we start launching some of our future
systems with no routed IP, is there a clean path for them to "migrate" 
and be full participants in the future network. More importantly, since
the lunar systems are new systems, does it make sense to build some
using legacy protocols, then have to upgrade them to support future
IP protocols, and also do long-term support of gateways to support
interaction between legacy and IP systems. Isn't it more beneficial
to take this opportunity to deploy a whole fleet of new systems and
start the first systems with the technologies we want to end up with.

   Starting new missions with IP technologies also means that they can
start benefiting from the greatly increased options for early testing
between various subsystems.  Systems don't need to wait until final
integration and test to start doing interoperability tests.  With
IP interfaces built in, systems can start running basic interface
and functionality tests while they are still at their manufacturing
locations.  This can be done years earlier than normal I&T and find
basic problems much earlier when they are easier and cheaper to fix.

   I agree that we don't need to pin down all the details now, but we do
need to have some sort of plan on how things will roll out.  There are
lots of options for starting with static IP routes, managed IP tunnels,
automated IP tunnels, and eventually move to whatever mobile ad hoc
networking mechanisms get developed by the internet community.  We also
need to start leveraging Internet security options, and all sorts of
other standard IP network capabilities and applications.

   We may not need all the network routing capabilities for 10 years
but there seem to be lots of benefits from starting to make use of
end-to-end Internet technologies now.

Keith


----------------------------------------------------------------------
   Keith Hogie                   e-mail: Keith.Hogie at gsfc.nasa.gov
   Computer Sciences Corp.       office: 301-794-2999  fax: 301-794-9480
   7700 Hubble Dr.
   Lanham-Seabrook, MD 20706  USA        301-286-3203 @ NASA/Goddard
----------------------------------------------------------------------