RDMA over the Internet Protocol Suite BOF (roi) Tuesday, March 19 at 1300-1400 ================================== CHAIRS: Stephen Bailey Allyn Romanow DESCRIPTION This BOF will consider the problem being addressed by the proposed RDMA (Remote Direct Memory Access) over IP working group, and also consider a charter for it. At high speeds (today 1 Gbits/s and higher), copying in network I/O typically causes high system costs in end-hosts. This problem is ultimately attributable to the high cost of memory bandwidth relative to both CPU and network bandwidths. The problem can be substantially improved using "copy avoidance." RDMA is a technology that allows the network adapter, under control of the application, to place data directly into and out of application buffers. This capability is also referred to as "direct data placement". Special purpose memory-to-memory fabrics have been developed and RDMA is widely used by databases and other networked applications sensitive to overhead. The relatively higher overhead of TCP/IP has so far prevented it from being similarly used as such an interconnect. High costs associated with copying is an issue primarily for large scale systems, typically for servers in data centers. An IP-based solution developed within the IETF is desirable for interoperability of various network fabrics. In addition it is especially important for the standardization to be directed by the IETF because interconnection technology will soon move out of a "back room" data center and start to be used over the wide area in the Internet. AGENDA Agenda Bashing Discussion of Problem Statement Discussion of Proposed Charter DOCS RDMA Over IP Problem Statement: http://ietf.org/internet-drafts/draft-romanow-rdma-over-ip-problem-statement-00.txt Proposed RDMA over IP (ROI) Working Group Charter: http://www.cs.uchicago.edu/~steph/roi-proposed-charter-00.txt