On 7/28/2016 1:40 PM, Olle E. Johansson wrote:
> ...
>> Those vendors shouldn't be looking at TCP options or TCP at all. It's
>> none of their business.
> Well, the question is what they are trying to solve by breaking TCP?
> I  personally suspect that the latency on some radio networks cause
> a lot of retransmits. By adding a TCP proxy they avoid retransmits 
> wasting bandwidth and at the same time causing issues when it 
> doesn’t work. Can we help them with a better solution if this is 
> the problem?
They should be doing FEC or ARQ at the link layer or via a tunnel across
multiple radio links, but that requires more effort on their part
(inserting and coordinating multiple components, rather than just one).

> Does anyone know what the problem they are trying to solve is?
> (Still trying to find information)
The one you cite above is one. Another intermittent connectivity.
Another is bandwidth or device capability, but in those cases the proxy
can be visible.
>> This issue cannot be fixed merely by reacting to what vendors deploy.
>>
>> The solution has been clear for a long time - *compliance verification*.
>> I assure you that vendors that get sued for saying "Internet compatible"
>> who are not would behave differently.
> So where do we have a good specification that we agree on defining
> “Internet compatible” ?

RFC1122/1123 for hosts.

RFC1812 for routers.

Joe


Reply via email to