On 2016-03-17 18:48, joel jaeggli wrote: > On 3/17/16 2:50 AM, Josh Howlett wrote: >>> The takeaway for me is that the end-to-end principle is dead. The >>> times where all intelligence should be in end hosts only, no >>> middleboxes, and KISS have gone, and techniques going in the >>> opposite direction are not only in use, but even embraced by IETF >>> operations. >> >> +1 >> >> But the fact that these solutions exist points to a need unmet by >> existing e2e approaches. Someone in IETF operations had presumably >> thought out their requirements when this service was procured. >> Embedding e2e within a solution that satisfies those same operational >> requirements sounds like a potential IETF WG charter in the making. > > I'm not really sure what you folks are talking about anymore. A CDN > whether internal or a third party is an extension of the applications > stack, not a bump in the wire. > > building scalable performant low latency services is a distributed > systems problem and there are a number of areas of the ietf activity > (all of them probably) involved in doing. very true
Attachment:
signature.asc
Description: OpenPGP digital signature