Dear authors,
I have some observations on the new map-dhcp draft:
Sec 1, para 2
Networks with numerous CEs require dynamic configuration of these
parameters on CEs, which forms the requirement for a dynamic
provisioning mechanism using DHCPv6 [RFC3315] options that are the
subject of this document.
[Qi] The dynamic feature of DHCPv6 is for the management of IPv6
address/prefix, rather than for that of IPv4 address/port-set. And the subject
of this document is, IMHO, assigning IPv4 address/port-set across IPv6 network
using DHCPv6 options. It's also not the DHCPv6 server's job to dynamically
manage the IPv4 address/port-set.
sec 3, bullet 1:
Key parameters of a BR include: i) The IPv4 Prefix - Used
to derive the CE's IPv4 address; ii) The Embedded Address bit
length - Used to derive how many, if any, of the CE's end-user
IPv6 prefix are to be mapped to its IPv4 address and port-set.
iii) The IPv6 prefix - used to scope the CE's IPv6 domain for the
S46 service
[Qi] All these are for MAP. What lw4o6 needs are IPv6 address/prefix, IPv4
address and port set (optionally). The IPv6 address/prefix is assigned through
normal DHCPv6 process, not through the newly defined option. So would it be
helpful that we reflect that in this document?
Sec 6, bullet 3:
(0/0 IPv4 prefix) + (128 IPv6 prefix) ==> encapsulation mode
(0/0 IPv4 prefix) + (shorter than 128 IPv6 prefix) ==> translation mode
[Qi] What happens if the length of the IPv4 prefix is zero, while the content
of IPv4 prefix is not 0.0.0.0?
My suggestion is once the length of the IPv4 prefix is zero, the CE SHOULD
ignore the content of the IPv4 prefix.
Sec 4.6, bullet 4
o The presence of a OPTION_S46_BRULE option indicates to the CE that
NAPT44 mode of operation is to be used, and the conveyed A+P
parameters applied.
[Qi] Does this mean no matter whether the CE is configured with a shared IPv4
address, NAT44 mode will be used anyway?
Sec 7:
This type of DHCP server and configuration store deployment
is recommended when the intent is to utilize full IPv4-IPv6 address
independence of any given CE, by issuing unique options to each CE.
[Qi] Port Parameters Option is actually a per-client option. This is true no
matter whether the full IPv4-IPv6 address independence is needed. Length of
PSID is a parameter shared by CEs in a domain. Maybe we need to clarify that?
More extensive than those possible by means of DHCPv6
in conjunction with the options in this document, should consider the
use of DHCPv4 over IPv6 as specified in
[I-D.ietf-dhc-dhcpv4-over-ipv6].
[Qi] I think there should be a reference to dhcpv4-over-dhcpv6.
Nits:
sec 1, para 2
responsible for forwarding between an
IPv6 network domain and external public IPv4 network,
=> an external public IPv4 network
S46 Port Parameters Option
=> sub-option? Not sure
OPTION_S46__PORTPARAMS => OPTION_S46_PORTPARAMS
Best Regards,
Qi
_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires
