(Apologies for cross-posting) Dear all,
We have uploaded a draft describing our proposed transport re-architecture, with its relevance and implications to multipath transport. Our hope is that this draft starts (or continues?) a conversation on a refactoring of the transport layer to accomodate new services and network elements. We also hope that this draft serves as a starting point for discussions about multipath transport and for an architectural framework within which to place multipath-tcp work at the IETF. ----------- A Next Generation Transport Services Architecture http://www.ietf.org/internet-drafts/draft-iyengar-ford-tng-00.txt Abstract: While there is substantial community interest in next-generation multipath-capable Internet transports, evolutionary pressures have gradually eroded the simplicity of the Internet's original transport architecture to a point where it is no longer realistically applicable to new tranports. This document proposes a new architectural framework for next-generation multipath-capable transport protocols, focusing immediately on multipath TCP but taking care to allow for generalization to other multipath-capable transports. The architecture places emphasis on enabling new multipath features in a safe, TCP-friendly, and backward-compatible fashion, retaining full interoperability with both existing applications and existing network infrastructure, and enabling reuse of existing protocols as much as possible while providing incremental deployment paths to new, more powerful and/or more efficient protocols. The architecture re-establishes the long-lost principles of end-to-end reliability and fate sharing, in the presence of existing and future network middleboxes, and enables the deployment of transport-neutral end-to-end protection without interfering with these policy-enforcing or performance-enhancing middleboxes. This document describes architecture goals, a layering model supporting these goals, abstract properties of the interfaces between the architecture's new layers, general approaches to multipath congestion control and how they fit into the architecture, realistic protocol design and incremental deployment paths, and ways in which this document complements and relates to ongoing protocol design activities in the IETF. ------------ Comments and thoughts welcome! - jana -- Janardhan Iyengar Assistant Professor, Computer Science Franklin & Marshall College http://www.fandm.edu/jiyengar
