[Sis-csi] No call today

[Marc Blanchet]

Le 06-12-21 à 14:22, Keith Hogie a écrit :

> Will,
>
>   You mentioned that you have been running this over Ethernet and are
> not sure what happens with serial links.  What about the process for
> generating link local addresses.
>
>   I think the normal process for generating a link local address is to
> use your MAC address in the last 64 bits.  This works great for  
> Ethernet
> but serial links don't have MAC addresses.  A MAC address is nice
> because they are globally administered and unique.  What do we do to
> avoid people randomly assigning their own serial link addresses and
> ending up with one that matches someone else.

IPv6 specs discuss this point (If I recall well, it is in RFC2461).  
One can use:
- a MAC address if available
- any other unique address
- randomly generated interface identifier
- even crypto-generated with a key (SEND)!

on PPP links (often used for serial communications and point-to-point  
links), one way (used in wireless world for example) is to have PPP  
negotiate the interface identifier (through the IPV6CP control  
protocol), so the interface identifier is actually given by the PPP  
peer so it is unique between the two peers that make the (serial  
point-to-point) link. [RFC2472]

ie. there are many ways to create link-local addresses and you have  
2^64 combinations possible.
DAD should be used to test the uniqueness.

Marc.

>
>   Just one of those differences between the Ethernet multi-drop LAN
> world and the serial point-to-point world.
>
> Keith Hogie
>
>
>
> Ivancic, William D. (GRC-RCN0) wrote:
>> Keith,
>>  I believe on can turn neighbor discovery off.  I believe many  
>> manets are
>> interested in turning neighbor discovery off to reduce chatter and
>> because one may not be connected to the manet or reachable during
>> another's neighbor discovery process.  Space applications would be
>> similar I suspect.
>> Yes, for any routing protocol, one needs bidirectional links.   
>> Otherwise
>> one has  to simply use static or default route.  These of course  
>> could
>> be a weighted route such that if no other dynamic routes exist,  
>> send to
>> the default.
>>  Regarding the NEMO comment:
>>                   /------/                  /---.--/
>>                     |                         |
>>   2001:DB8:2:2::/64 |                         | 2001:DB8:3:3::1/64
>>                     |                         |
>>               +-----+----+              +-----'----+
>>               | Router A |..............| Router B |
>>               +----------+              +----------+
>>   Example 1           10.1.2.1       10.1.2.2
>>   Example 2        10.100.20.1       10.200.50.1
>>   Example 3    2001:DB8:1:1::1       2001:DB8:1:2::1
>> Example 1 is how IPv4 routing has to take place  (assume class C /24
>> networks)
>> Example 2 will not allow for IPv4 routing (RIP/OSPF) as the  
>> interfaces
>> are in different sub-networks.
>> Example 3 works for IPv6 because of link-local addressing.  We had  
>> three
>> routers in series, not just the two shown.  Okey will be writing  
>> this up
>> after the Holidays.  Everything so far has been over Ethernet.  I  
>> want
>> to make sure the same goes for Serial.  I believe is should, but I
>> always test my beliefs and have often been surprised.   Will
> ----------------------------------------------------------------------
>   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
> ----------------------------------------------------------------------
>
> _______________________________________________
> Sis-CSI mailing list
> Sis-CSI at mailman.ccsds.org
> http://mailman.ccsds.org/cgi-bin/mailman/listinfo/sis-csi