[Sis-csi] CANDOS networking details

[Ivancic, William D. (GRC-RCN0)]

________________________________

	From: sis-csi-bounces at mailman.ccsds.org
[mailto:sis-csi-bounces at mailman.ccsds.org] On Behalf Of Adrian J. Hooke
	Sent: Wednesday, February 07, 2007 10:30 AM
	To: Cislunar Space Internetworking
	Subject: RE: [Sis-csi] CANDOS networking details
	
	
	At 05:02 AM 2/7/2007, Ivancic, William D. \(GRC-RCN0\) wrote:
	

		#2.  Static routing does exactly what you tell it to do.
Static routes tend to default to the highest priority route (although
this is configurable in Cisco routers and probably other systems). 


	  
	

Adrian,

>"Probably"? Don't we know the full range of capabilities offered by
other manufacturers? 
>Are we sure that we aren't making assumptions that lock us into a
single vendor?

"Probably" is used because I don't have the dollars or staff to perform
test with other equipment.  Believe me, we would like to see what Linux
and BSD boxes would do as well as Vxworks, and Juniper and Nortel,
Hitachi, Fujitsu, Procket and whatever the China router manufacture.
And ... This should be done by someone.

There are many other question regarding how other manufactures handle
things as well.  For example, I have not figure out if one can trick a
Cisco box into setting up the same subnet or overlapping subnets on two
or more interfaces - particularly regarding IPv4.  However, some other
manufacturers may allow this.  I suspect a lot of this is implementation
specific and a lot of the "not allowed" is because it really doesn't
make sense to allow such a configuration in terrestial operational
systems.  By allowing users to misconfigure networks just results in
calls to the technical department to fix poor design.  So...it make more
sense to not provide the bullets for someone to shoot themselves in the
foot with (from a manufactures and equipment providers point of view).  

> And if we have to set up every route by hand, what are we gaining over
what we have now?

If we use static routing and no dynamic routing, we gain much less than
we should.  I hesitate to say we gain nothing.  If we are just speaking
of CEV replacement for STS to get to ISS then we gain little.  If we
look at multi-hop, we probably gain more.   End-to-End Security should
be easier even with static routes.   

The main point, however, is that the technology is there today in IPv6
to do dynamic routing easily, so why implement a more difficult system
as is proposed - static routing.  Static routing (or layer-two
switching) will only cost more and lead to misconfigurations -
particularly as the system gets more complex.  

Will

Will