Giving applications access to Gb/s networking

Abstract

Network fabrics with Gigabit per second (Gbps) bandwidths are available today, but these bandwidths are not yet available to applications. The difficulties lie in the hardware and software architecture through which application data travels between the network and host memory. The hardware portion of the architecture is often called a host interface and the remainder of the protocol stack is implemented in host software.In this paper, we outline a variety of approaches to the architecture of such systems, examine several design points, and study one example in detail. The detailed example, an ATM Host Interface and Operating System support built at the University of Pennsylvania, illustrates design tradeoffs for hardware and software, and some of the implications of these tradeoffs on applications performance.

Accelerating application-level security protocols

“…Our cost-breakdown for the various costs, summarized in Figure 1, shows that over 20% of the processing time is spent copying data over these two buses. Such findings have been reported in the past for other systems [10], [11], [12], [13], and are by no means limited to cryptography or network I/O.…”

“…[15] began examining cryptographic subsystem issues in the context of securing high-speed networks, and observed that the busattached cards would be limited by bus-sharing with a network adapter on systems with a single I/O bus. A second issue pointed out in that time frame [16] was the cost of system calls, and a third [10], [11], [12], [13] the cost of buffer copying. These issues are still with us, and continue to require aggressive design to reduce their impacts.…”

The OKE Corral: Code Organisation and Reconfiguration at Runtime Using Active Linking

“…First, such numbers often say more about the traffic capture (e.g. polling or interruptdriven) than about the processing [ST93]. Second, we are really interested in how the Corral compares to typical ANs and this concerns primarily the nature of code execution: in-kernel native code versus interpreted code in userspace.…”

Open Packet Monitoring on FLAME: Safety, Performance, and Applications

“…The original FLAME architecture interacted with the network interface exclusively through interrupts. As others have noted [15,22], under high rates of incoming network traffic, interrupt handling can degrade performance. More recent versions of FLAME poll the network interface card (NIC) to read packets to avoid performance degradation.…”