Re: [Softwires] The port mapping issue

[Tom Taylor <[e-mail]>]

On 27/03/2013 4:09 PM, Wojciech Dec wrote:
> On 27 March 2013 16:59, Tom Taylor <[e-mail]> wrote:
>> The meeting minutes record a disagreement over what port mapping algorithm
>> to use. This affects both MAP-E and LW 4over6. As I understand it:
>>
>> - either of these two technologies will work with either contiguous ports or
>> ports scattered according to the GMA algorithm
>>
>> - the real objection to GMA comes from Alain Durand, who wants to set up
>> simple min-port, max-port filters on his network equipment.
>>
>>
>> We all agree that port scattering offers negligible security advantage.
>>
>> The reason that I heard given for preferring GMA for MAP-E is that it
>> eliminates a restriction on the End-User Ipv6 address because the PSID is
>> free to range from 0 upwards rather than from some higher number upwards. I
>> don't follow this argument for two reasons:
>>
>> - you now have a restriction that the offset field A must range from 1
>> upwards
>>
>> - the PSID field has an upper limit 2^k-1 imposed by the sharing ratio,
>> imposing a further restriction on the End-User IPv6 address value.
>>
>> Could someone spell out more clearly why the GMA was seen as necessary for
>> MAP-E?
>
> GMA codifies (power of 2) logic that translates easily into hardware
> based longest match lookups. Furthermore, it is actually trivially
> simple to implement on any platform (not hardware restricted thus):
> Something like 2 lines of C code, in optimized form. Lastly, it is
> very efficient in terms of info encoding: a PSID conveys a code-point
> that maps the entire port space or port range. Finally, embedding a
> explicit port-range in an IPv6 address is not an option (the N:1 rule
> case).
>
> Regards,
> Woj.

Thanks, Woj. I think I started off in the wrong direction, and should 
focus specifically on Suresh's proposal: GMA for both Light-Weight 4 
over 6 and MAP-E, with default a=0 for the former and a=6 for the latter.

When a=0, the GMA algorithm does degenerate to assigning ports as a 
single contiguous range per CPE. Min_port for a given PSID is given by 
the formula PSID * range size, or, in the notation of MAP-E section 5.1, 
PSID * 2^m. Max_port is given by Min_port + range size - 1.

With MAP-E, range size is inferred indirectly from the combination of 
the IPv4 prefix length in the Basic Map Rule and the number of EA bits. 
This information is not available for Light-Weight 4 over 6. Hence I 
believe the proposal to use GMA with a=0 for the latter amounts to 
explicit provisioning of PSID and range size to both the BR and the CPE.