I believe something missed in these discussions is that pertaining to "the data-center". A data-center is not a technology, it is a grouping of assets. Those

I believe something missed in these discussions is that pertaining to "the data-center". A data-center is not a technology, it is a grouping of assets. Those assets and their associated services need to communicate not just with each other, but with other assets and services on the internet. Regardless of what incredible technical solution one creates, it will have to allow for simultaneous existence of current IP protocols in parallel with whatever proposed replacements to exist seamlessly with one another or significant adoption would never occur. A data-center is not an isolated bubble, at least not any more unless one wants to translate said protocols through a single point of success gateway. Should such a replacement ever occur it will have to be done piece by piece until there is nothing left using current IP protocols. So I believe people should all create their proposed protocols and give businesses a low-friction path to adoption one service at a time. As more applications adopt said protocols, the most popular, highest reliable, most performant least friction path will likely win and if successful then at some distant point in the future perhaps most existing IP protocols could be deprecated. As a reminder, each application will have to adopt libraries to speak on this protocol and know how to utilize it. There will be a "battle hardening" period to work out the bugs and security controls. All of the network gear in the entire path between data-centers and clients will need OS/firmware/Asic updates to understand this protocol. Given the transition speed to IPv6 as an example, this could be a very long road. There is also some discussion of QUIC and SPDY. Those are not new IP protocols. Those are new standards within an existing L7 application protocol HTTP that still utilize existing L3/L4 protocols. Replacing TCP and UDP? in the data-center means a new protocol in /etc/protocols not encapsulated in an existing protol such as protol 6 TCP or protocol 17 UDP. The network gear and OS on every device in the path will need to understand this new protocol. Tools such as tcpdump and libraries such as libpcap would need to be updated to understand these new protocols before they would even be used in a development environment. Could it be that I misunderstood the intent and perhaps we just want yet another new L7 application protocol on top of UDP?