The iSCSI protocol is the IETF standard that maps the SCSI family of application protocols onto TCP/IP enabling convergence of storage traffic on to standard TCP/IP fabrics. The ability to efficiently transfer and place the data on TCP/IP networks is crucial for this convergence of the storage traffic. The iWARP protocol suite provides Remote Direct Memory Access (RDMA) semantics over TCP/IP networks and enables efficient memory-to-memory data transfers over an IP fabric. This paper studies the design process of iSCSI Extensions for RDMA (iSER), a protocol that maps the iSCSI protocol over the iWARP protocol suite. As part of this study, this paper shows how iSER enables efficient data movement for iSCSI using generic RDMA hardware and then presents a discussion of the iWARP architectural features that were conceived during the iSER design. These features potentially enable highly efficient realizations of other I/O protocols as well.

No abstract

The original RFCs that specified the Remote Direct Data Placement (RDDP) protocol suite did not create IANA registries for RDDP error codes, operation codes, and function codes. Extensions to the RDDP protocols now require these registries to be created. This memo creates the RDDP registries, populates them with values defined in the original RDDP RFCs, and provides guidance to IANA for future assignment of code points within these registries.

User space applications must invoke kernels calls in order to have the Operating System handle file system and device driver processing for storage requests. There is a high overhead associated with such calls, including the processing of suspending, scheduling and dispatching threads, interrupts, cache misses, etc. The overhead contributes to wasted processor cycles because all work on the thread being switched is stopped until the task switch is complete. The problem is worse for applications with a high storage I/O to compute ratio. Alternatives to enable user space applications to pass storage requests directly to I/O adapters without run-time involvement from the Operating System would eliminate this overhead .