OMG!
Typo below - NEAT is the a research project implementing TAPS, and I wrote this 
email after the social… enough said…

very sorry: I meant to say: please discuss on the TAPS mailing list.


> On 19. jul. 2016, at 23.05, Michael Welzl <[email protected]> wrote:
> 
> [ Please discuss on the NEAT mailing list ]
> 
> 
> Hi,
> 
> The TAPS WG is trying to identify which services transport protocols offer, 
> and then identify which of these services should be exposed to an application 
> that tries to communicate while being agnostic to the transport protocol 
> below. A TAPS system should then make a choice about the right transport 
> protocol and configure it autonomously, and implement e.g. protocol 
> fall-backs and such.
> 
> Some transport services must be exposed or else they could never be used 
> (e.g. unordered message delivery). Some can optimize performance if they are 
> exposed and cannot be used without application involvement, but won’t make an 
> application fail if they are not used (e.g. turning the Nagle algorithm 
> on/off). The remaining ones could be “automatized”, i.e. a TAPS protocol 
> could potentially make a decision on its own about them.
> 
> Note that this raises the abstraction level of the API applications use to 
> talk to the network. As always, this comes with benefits (of automatizing 
> more things, and not requiring one specific transport protocol, ..), but also 
> with reduced control over the available resources. As an extreme example, 
> nobody in their right mind would write wireshark over TAPS  :-)
> 
> One proposal is to consider everything related to the usage of multiple paths 
> automatable. That is, not expose it as a TAPS service.
> 
> What do people here think about this?
> 
> Cheers,
> Michael
> 
> _______________________________________________
> Taps mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/taps

_______________________________________________
Taps mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/taps

Reply via email to