[Sis-csi] IPv6?

[Lloyd Wood]

At Saturday 09/06/2007 11:05 +0200, Chris.Taylor at esa.int wrote:
>I am well aware of the emerging initiatives in DTN and the fact that by
>introducing another layer of bundling we could resolve some of the issues
>with IP. I'm also well aware that we can fly a commercial router just for the
>benefit of sending point to point traffic from a payload to ground. Something
>we can do anytime with existing software and at zero cost. But this is not my
>point. I'm asking what we really need in the next 15 year or so period, a
>period in which the commercial terrestrial standards will evolve out of all
>recognition and many (most) of the standards used today will have long gone,
>as will the commercial support and components needed for long duration
>missions.  Just how many IP addresses are we going to need for the flight
>infrastructure in place at the time - 4, 5? Wow. Do we really need a fully IP
>routed flight architecture to support a few IP cognizant nodes.

IP routing isn't needed onboard for exploration, except on manned missions - the point section 14 of the paper I pointed out makes with respect to scheduling, ownership, and contention. (Experiments with our router in orbit are unrelated to that particular point. The router fits into an already existing IP infrastructure; the infrastructure doesn't require the onboard router. You'll note that I didn't mention the onboard router in the previous mail.)

Saratoga and other similar UDP protocols do not require the layer of bundling to function in the face of disruption or delay, as those papers make clear.

The SSTL satellites don't use onboard routing. But they do contain multiple IP nodes, and use multiple IP addresses. There's your realistic configuration and deployment of IP in space, right there. I fail to see how the Disaster Monitoring Constellation (with five satellites launched, three pending) is 'commercially driven nonsense', and the Disasters Charter would not agree with you. The model used on the DMC satellites can be applicable to deep-space exploration. Don't confuse that model with an experimental onboard router.

I also fail to see how terrestrial standards will evolve out of all recognition in fifteen years. IP is over thirty years old, as is Ethernet. Frame relay is over fifteen years old.

What do you see as your existing problems?

L.



>I would contend that the recent pushing of IP in the context of exploration
>is essentially hype. Dangerous hype at that, as it obscures the real issues
>that need to be tackled to resolve existing problems and agree an
>evolutionary path forwards. Until I see a realistic configuration (not the
>futuristic, commercially driven nonsense) of what we can expect in 15-20
>years, and set of accompanying comms requirements that justify the need for
>IP, I will continue to doubt the goals, objectives and understanding of the
>IP exponents.
>
>//ct.
>
>
>                                                                             
>             Lloyd Wood                                                      
>             <L.Wood at surrey.ac.                                              
>             uk>                                                          To 
>             Sent by:                   Chris.Taylor at esa.int                 
>             sis-csi-bounces at ma                                           cc 
>             ilman.ccsds.org            CCSDS Cislunar Space Internetworking 
>                                        WG <sis-csi at mailman.ccsds.org>       
>                                                                     Subject 
>             08/06/2007 19:56           RE: [Sis-csi] IPv6?                  
>                                                                             
>                                                                             
>                                                                             
>                                                                             
>                                                                             
>                                                                             
>
>
>
>
>At Friday 08/06/2007 17:45 +0200, Chris.Taylor at esa.int wrote:
>>Just to stir things up a bit, has it really been decided that we will use IP
>>on our future links. From the discussion it seems like a done deal but our
>>ESA studies and opinion is that IP doesn't bring us much other than a bit
>>more address space that we probably don't need anyway. Rather than  discuss
>>the merits of IPv4 and 6 it may be more productive to critically examine the
>>application of IP to see how it may be employed or not. I should say that I
>>have no particular issue with the use of IP its just that I think the
>problem
>>is much wider and by concentrating on v4/v6 there is a danger of missing the
>>real problems - IP doesn't work on links that have disjoint connectivity
>
>Actually, IP works just fine on links with disjoint connectivity, as large
>amounts of delay/disruption tolerant networking work and mobile ad-hoc work
>by many people have shown. For example, we've developed our own IP/UDP-based
>transport protocol for moving files over disjoint links, and there are other
>protocols that work over IP in these environments. TCP won't work well, but
>then TCP's operational range is surprisingly limited. TCP is not IP.
>
>See:
>http://www.ee.surrey.ac.uk/Personal/L.Wood/dtn/
>ftp://ftp-eng.cisco.com/lwood/dtn/README.html (IE users: passive ftp must be
>ON in Tools/Options... Advanced tab)
>- Saratoga: a Delay-Tolerant Networking convergence layer with efficient link
>utilization, Lloyd Wood, Wesley M. Eddy, Will Ivancic, Jim McKim and Chris
>Jackson, submitted to the Third International Workshop on Satellite and Space
>Communications (IWSSC '07), September 2007.
>- Saratoga: A Convergence Layer for Delay Tolerant Networking, Lloyd Wood,
>Wesley M. Eddy, Will Ivancic, Jim McKim and Chris Jackson, work in progress
>as an internet draft, version -00 submitted to the IETF, May 2007.
>
>For a detailed discussion of how IP can be used in the disjoint environment
>of intermittent space links, see:
>http://www.ee.surrey.ac.uk/Personal/L.Wood/cleo/
>ftp://ftp-eng.cisco.com/lwood/cleo/README.html (IE users: passive ftp must be
>ON in Tools/Options... Advanced tab)
>- Using Internet nodes and routers onboard satellites, Lloyd Wood, Will
>Ivancic, Dave Hodgson, Eric Miller, Brett Conner, Scott Lynch, Chris Jackson,
>Alex da Silva Curiel, Dave Cooke, Dan Shell, Jon Walke and Dave Stewart,
>special issue on Space Networks, International Journal of Satellite
>Communications and Networking, vol. 25 issue 2, pp. 195-216, March/April 2007
>- particularly section 14.
>
>The primary argument for using IP imo is that you can just reuse
>well-developed commercial technologies without reinventing the wheel,
>minimising your overall software development and testing costs.
>
>(How much is esa spending on developing Spacewire? IP already runs well over
>LVDS HDLC serial links, and is used on them in space.)
>
>regards,
>
>L.
>
>
><http://www.ee.surrey.ac.uk/Personal/L.Wood/><L.Wood at surrey.ac.uk>
>
>_______________________________________________
>Sis-CSI mailing list
>Sis-CSI at mailman.ccsds.org
>http://mailman.ccsds.org/cgi-bin/mailman/listinfo/sis-csi