[alto] Discussion about Extended ALTO Endpoint Address

[Jensen Zhang <[e-mail]>]

Hi all,

Because several recent drafts (cellular addresses and fcs) meet the same
requirement of registering new ALTO address types, I think it's time to
propose it as a general problem and have a good discussion.

If you take a quick look at the section 5.2.2 of the latest fcs revision
[1], you will find we are requiring a new registry called "ALTO Address
Type Conflict". The key insight is that some address types cannot appear in
the associated source and destination addresses. For example, the following
filter is invalid:

{
  "srcs": [ "ipv4:192.168.1.100" ],
  "dsts": [ "ipv6:::1" ] // Just take an example. The loopback address
should never appear in the ALTO resource.
}

In the legacy ALTO, we can just declare different address types conflict
with each other. But when we register a new address type A, it may be
allowed to appear with some existing address types. So we need to indicate
which existing address types conflict with the new address type A, and
which ones do not.

"ALTO Address Type Conflict Registry" [2] is a solution proposed by us. But
there is a potential issue: it is not scalable. Because the future
registered address type will have to declare more and more conflicts as the
growth of the address type registry.

And since it is a general problem, do you think it is a good idea to start
a new draft focusing on this problem? I think both cellular addresses draft
and the endpoint address extension part of FCS will benefit from it.

[1] https://tools.ietf.org/html/draft-gao-alto-fcs-04#section-5.2.2
[2] https://tools.ietf.org/html/draft-gao-alto-fcs-04#section-9.2

Thanks,
Jensen