[Taps] review: draft-grinnemo-taps-he-02

["Bajpai, Vaibhav" <[e-mail]>]

>   Title:	  Happy Eyeballs for Transport Selection
>   URL:            https://www.ietf.org/internet-drafts/draft-grinnemo-taps-he-02.txt

Overall, I like this document (suggestions for improvements below). I hope an
HE mechanism helps get future transport protocols get wider adoption.

-- Vaibhav


  To be compliant with RFC 6555 [RFC6555], the Policy Management component
  SHOULD, in those cases there are no policies telling otherwise, give priority
  to IPv6 over IPv4.

-- RFC 6555 does not give priority to v6. Yes, this was the original
motivation for the document, but RFC 6555 states that the priority is dictated
by the address selection policy [RFC 6724]. In majority of the scenarios this
boils down to giving priority to native v6. However, if a host has native v4
and Teredo connectivity, a HE implementation will happily give priority to
native v4 over Teredo.



  To minimize the number of connection attempts that are initiated,
  the Transport Probing component SHOULD cache the outcome of
  connection attempts in a repository kept by the Policy Management
  component.

  Cached connection-attempt results SHOULD be valid for a configurable
  amount of time after which they SHOULD expire and have to be repeated.

-- How long should this cache be? How is this configured?


  D then governs the delay between the initiation of the connection
  attempts C1 and C2.

-- how is delay D determined?




  For the Transport Probing component to be able to efficiently use the
  connection-attempt cache, already-initiated, non-winning transport solutions
  SHOULD NOT be terminated as soon as a winning connection has been established.

-- I found this sentence confusing. Perhaps an example helps here.


  As pointed out in RFC 6555 [RFC6555], a HE algorithm should not waste
  networking resources by routinely making simultaneous connection
  attempts.  To this end, the HE algorithm should cache the outcome of
  previous connection attempts to the same peer.  The impact and
  efficiency of the HE algorithm has been evaluated in
  [Papastergiou16].  The paper suggests that caching significantly
  reduces the CPU load imposed by a HE mechanism.


-- Perhaps make it clear that [Papastergiou16] discusses the impact of memory
   and cpu resources on the end-host and not networking resources on the wire.



  Consider a scenario in which an IPv4-only client using the ...

-- Why not v6-only client instead? :)



-- It was unclear to me how the the list of candidate protocols is generated
   by the policy manager. An example would help here. I am even thinking
   perhaps this is out of scope of this document.


--------------------------
Vaibhav Bajpai
www.vaibhavbajpai.com

Postdoctoral Researcher
TU Munich, Germany
--------------------------