m**t
发帖数: 1292 | 1 note this RFC is not strictly speaking a NAT66, indeed the author has
changed its naming to NPTv6 if you've noticed and it is also experimental.
None of other nat66 or napt66 drafts has survived
I think the only potential legitimate reason is as what this NPTv6 claims to
solve, but don't think there is any mandatory cases..meaning you can always
configure/modify/plan/enforce your network addressing scheme without using
the NPTv6 solution although it provides convenience.
What is a compelling r... 阅读全帖 |
|
f*****m
发帖数: 416 | 2 IPv6也需要NAT.
先不说NAT46 或者NAT64. 现在NAT66也出来了. |
|
|
f*****m
发帖数: 416 | 4 rfc6296
貌似不少客户对这个感兴趣,不过产品上好象还没有 |
|
z**r
发帖数: 17771 | 5 大家一般为什么会用这个？俺的情况是比较特殊所以需要这个 |
|
|
|
f*****m
发帖数: 416 | 8 IPv6为了scale, 路由要一路aggregate,这样核心路由表才能小.这样客户的路由就是
provider dependent的.问题是当年讨论这个的时候,只注意了scalability,没注意到客
户multi-homing的需要.客户如果multi-homing的话,就要多个PD的地址段,每台host也
会有多个unicast address (PD). 多个PD又带来诸多问题, load sharing, 换SP的时候
renumbering...Shim6历经多年才变成RFC,不过修成后也没谁用.
后来有开始用PI的地址段,这样换provider的时候不用renumber, 可是PI的后果是
scalability被破坏. 再后来LISP出来了,不过持怀疑态度的一大堆.再后来就是这个了.
.. |
|
z**r
发帖数: 17771 | 9 讲得很好，多谢。
每台Host多个unicast address的情况下，host如何来决定用哪个IPv6 address？
了. |
|
|
