Your take is really interesting, Patrik, and exactly the kind of knowledge I think RIPE and the broader technical community could inject into these processes. Would it make sense to do a quick write-up, explaining the technical difficulties/impossibilities of implementing what is currently (vaguely) defined in the draft, and requesting clarity and technical specifics? On Wed, Mar 19, 2014 at 7:44 PM, Patrik Fältström <paf@frobbit.se> wrote:
Agree, but that VPN can not be delivered (over the same IP based network) as internet access without the internet access being degraded. That is what I read the text say. And my point is that what this results in is that the customer of the internet access should continue to get whatever service they bought, irrespectively if some VPN service or whatever is transported in the same shared physical medium, L2 or L3 network.
If that is what the intention is, why do they not write that?
Patrik
On 2014-03-20 00:30, Innocenzo Genna wrote:
In my opinion, that kind of specialized services are a VPN. It’s no Internet.
----------------------------------------- Innocenzo Genna *Genna Cabinet Sprl * 1050 Bruxelles - Belgium
Skype: innonews Twitter:@InnoGenna Email: inno@innogenna.it <mailto:inno@innogenna.it>
my blog:http://radiobruxelleslibera.wordpress.com/ <http://radiobruxelleslibera.wordpress.com/> my music: www.innocenzogenna.com <http://www.innocenzogenna.com>
Il giorno 20/mar/2014, alle ore 00:03, Patrik Fältström <paf@frobbit.se <mailto:paf@frobbit.se>> ha scritto:
On 2014-03-19 20:13, Gordon Lennox wrote:
On 19 Mar, 2014, at 18:34, Innocenzo Genna <inno@innogenna.it <mailto:inno@innogenna.it> <mailto:inno@innogenna.it>> wrote:
15) “specialized service” means an electronic communications service */optimized for /*specific content, applications or services, or a combination thereof, */provided over logically distinct capacity and relying on strict admission control from end to end/*. It is not marketed or */usable/* as a substitute for internet access service; [its application layer is not functionally identical to services and applications available over the public internet access service;]
And that, particularly if the specialised service uses IP, is the problem?
And end-to-end means to a particular device or, more probably, an end network controlled by the service supplier.
I stopped liking "end-to-end" sometime back.
I have no idea what and how to implement technically what they talk about as "specialices service that does not impcat...".
In a packet based network, if the outgoing interface is not full, all packets will be forwarded as soon as possible.
If the outgoing interface is full, then one can either queue all packets equally (M/M/1 queuing theory) or one can have multiple queues (M/M/N). If one have a specialized service that have some special treatment, then by definition that implies longer delay on other queues (as packets get reordered).
Now, there are some special cases as well where the _services_ sold can be different (i.e. some business connection with some SLA that is higher than some SLA for end users paying less).
What I think is sad is that they did not stop at saying for example:
- Each provider of a service is required to always deliver to their customers the service they have promised to deliver. (Regardless of what other services they deliver to other customers on the same network...)
Not any silly end-to-end. No silly "specialized service" etc.
Then in other paragraphs they already (if I remember correctly) have wording about equal treatment, dominant provider of services etc.
Patrik
-- Meredith Whittaker Program Manager, Google Research Google NYC