Re: [Spud] SPUD Scope?

On Mon, Jun 8, 2015 at 7:26 AM, Daniel Kahn Gillmor
<dkg@fifthhorseman.net> wrote:
> Hi Mirja--
>
> On Mon 2015-06-08 03:57:28 -0400, Mirja Kühlewind wrote:
>> It’s not the different network providers that need different
>> information but you may want to provide different information about
>> different ‚kind‘ of services so that the provider might be able to
>> treat your traffic more appropriately (which effectively can provide
>> you a better service as well as save network resources). However, a
>> user might choose to not reveal this information and only get the
>> default treatment which might or might not be good enough (whatever
>> ‚good‘ means here). And here there might actually be differences
>> between the providers (and the contract that the user made with its
>> provider).
>
> We've been talking about generic middleboxes so far, and this is the
> Internet; almost by definition there will be middleboxes that have no
> contracts with the user.  It's quite likely for any given path between
> two peers that there will be *no* middleboxes on the path that actually
> have a contractual relationship with either peer.
>
Unfortunately this is isn't completely true between end users and
middleboxes. There are growing use cases where middleboxes are used to
enforce a service contract with users as to what applications are
allowed to pass through the network. For instance, some network
providers will not allow video as a basic service (IMO this is the
antithesis of net neutrality, but nevertheless we need to consider
this reality). One upshot of this is that middleboxes need to perform
DPI which becomes a problem if middleboxes can't identify the
application, so a consequence may be that e2e security (e.g. TLS) is
completely disallowed.

Tom


>      --dkg
>
> _______________________________________________
> Spud mailing list
> Spud@ietf.org
> https://www.ietf.org/mailman/listinfo/spud