Re: unicast address with non 64 bit prefix

2011-07-16 Thread Karl Auer
On Sun, 2011-07-17 at 10:45 +0800, Washam Fan wrote: > in our current design, the customer is allowed to set prefix-len to > non-64 value like 65. But according to discussions in this thread, it > seems to me we SHOULD NOT allow our customers to set prefix-len to > non-64 value unless they assign a

Re: unicast address with non 64 bit prefix

2011-07-16 Thread Washam Fan
Hi Kerry, Yes, they can be derived from PIO. But is there any possibility customers manually set subnet prefix? Now, we disable SLAAC on the appliance (due to it provides important services like web proxy/cache, web filtering, traffic monitoring) and use the 3 parameters to configure IPv6 address

Re: unicast address with non 64 bit prefix

2011-07-16 Thread Washam Fan
THanks Kerry and Brian, Now we provide 3 parameters to our customers to configure a IPv6 address manually: 128bit address, subnet prefix and default gateway (optional, if it is provided, it should be within the same subnet with the 128bit address). Should we set subnet prefix to a fixed value 64 i

Re: unicast address with non 64 bit prefix

2011-07-15 Thread Kerry Lynn
Brian, you're quite right. I had in mind only one particular class of IID. Washam, see Appendix A of RFC 4291 regarding modified EUI-64 IIDs with "u"=0. Also, RFC 3041 has been obsoleted; see RFC 4941 and its errata if privacy issues apply in your situation. -K- On Fri, Jul 15, 2011 at 6:25 P

Re: unicast address with non 64 bit prefix

2011-07-15 Thread Brian E Carpenter
The authors of RFC 4291 chose not to use the upper case normative keyword convention defined in RFC 2119. Therefore, 'required' means what in means in plain English (i.e. REQUIRED in RFC2119ish). Brian On 2011-07-15 19:03, Washam Fan wrote: > Hi, > > I quoted this paragraph from RFC4291: > >

Re: unicast address with non 64 bit prefix

2011-07-15 Thread Kerry Lynn
See also section 2.5.4 of RFC 4291: "All Global Unicast addresses other than those that start with binary 000 have a 64-bit interface ID field (i.e., n + m = 64), formatted as described in Section 2.5.1." So I'd say the answer is the IID MUST be 64 bits long, and MUST satisfy the properties of u

unicast address with non 64 bit prefix

2011-07-15 Thread Washam Fan
Hi, I quoted this paragraph from RFC4291: For all unicast addresses, except those that start with the binary value 000, Interface IDs are required to be 64 bits long and to be constructed in Modified EUI-64 format. But I am not sure how to interpret the text in a correct way. If I ass