This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).
The Internet has traditionally combined many orthogonal functions into transport protocols, creating significant technical and administrative hurdles to transport service evolution. New or specialized transport protocols are now nearly undeployable because they cannot traverse middleboxes such as NATs, firewalls, performance enhancing proxies, which have mushroomed in the past two decades; new congestion control schemes are restricted by the requirement to compete ?fairly? against traditional TCP flows; and deploying new features such as multi-homing is difficult because applications must be adapted to new naming and communication models.
Tng ("Transport next generation") is a new but incrementally deployable transport architecture that breaks the above evolutionary impasse by modularizing the transport layer. Tng breaks transports into four explicit layers - Endpoint Naming, Flow Regulation, Identity/Security, and Semantics - plus a cross-layer Negotiation service. By separating the network-oriented functions of endpoint naming and flow regulation from application-oriented transport semantics, Tng enables middleboxes in the network to enforce network policies and optimize flow performance cleanly across diverse network technologies and administrative domains, without interfering with end-to-end semantics. Tng's identity/security layer in turn enforces this separation between network- and appliation-oriented functions, avoiding past conflicts between middleboxes and IPsec.
By developing a working prototype and analyzing its performance and adaptability across a variety of real and simulated network environments, we expect that Tng will prove an important step towards breaking long-standing deadlocks between operators, new network technologies, and end-users.
