[openib-general] Re: IP addressing on InfiniBand networks (Caitlin Bestler)

[David M. Brean]

The IB spec says "A GID is a valid 128-bit IPv6 address (per RFC 2373) 
with additional properties / restrictions defined within IBA to 
facilitate efficient discovery, communication, and routing.", so I don't 
think it needs to say much more.  [Perhaps the IB specification should 
refer to the more recent RFC 3513 that obsoletes RFC 2373...]  One of 
those IBA restrictions result from being limited to assigning one 64-bit 
value for the upper 64-bits of the GID, called the GID prefix, per IB 
subnet.  If GIDs are used to represent IPv6 addresses on an IB subnet, 
then only one IP subnet prefix can be mapped onto an IB subnet.  This 
limitation goes along with the additional work of integrating GID 
management into the IP management applications for assigning IP addresses.

Restrictions on IP address assignment are removed when using the GID as 
specified by the IETF IPoIB as a link layer address and IB partitions 
can be used to create multiple IP links per IB subnet.  Also, the IP 
management application integration effort is significantly smaller.

Therefore, I recommend that a solution for mapping between IP and GIDs 
not rely on GIDs being used as the IP addresses.

-David

Michael Krause wrote:

> At 10:49 AM 6/30/2005, Roland Dreier wrote:
>
>>     Michael> Being the person who led the addressing definition for
>>     Michael> IB, I can state quite clearly that GID are NOT IPv6
>>     Michael> addresses.  They were intentionally defined to have a
>>     Michael> similar look-n-feel since they were derived in large part
>>     Michael> from Future I/O which had them as real IPv6 addresses.
>>     Michael> But again, they are NOT IPv6 addresses.
>>
>> The IBA spec seems to have a different idea.  In fact chapter 4 says:
>>
>>     "A GID is a valid 128-bit IPv6 address (per RFC 2373)...."
>
>
> I wrote the original spec here.  The text was supposed to be updated 
> to clarify that the rest of the sentence, i.e. with additional rules, 
> etc. thus making it not a real IPv6 address from the IETF's 
> perspective but something quite close.  The intention was to allow one 
> to manage the fabric by having mapping functions from traditional IP 
> management applications to IB GID to minimize the amount of work to 
> enable IB within a solution. 
>