The power of batching in the Click modular router

Abstract

The Click modular router has been one of the most popular software router platforms for rapid prototyping and new protocol development. Unfortunately, its internal architecture has not caught up with recent hardware advancements, and the performance remains sub-optimal in high-speed networks despite its benefit of flexible module composition.In this work, we identify the performance bottlenecks of the existing Click router and extend it to scale with modern computer systems. Our improvements focus on both I/O and computation batching, and include various optimizations for multi-core systems and multi-queue network cards. We find that these techniques improve the performance by almost a factor of 10, and the maximum throughput reaches 28 Gbps of minimum-sized IPv4 packet forwarding speed on a single machine.

Fast userspace packet processing

“…However, it only supports the in-kernel version of Click. DoubleClick [11] focuses on batching to improve overall performances of Click with PacketShader I/O [8]. SNAP [23] also proposed a general framework to build GPU-accelerated software network applications around Click.…”

Research Challenges for Network Function Virtualization - Re-Architecting Middlebox for High Performance and Efficient, Elastic and Resilient Platform to Create New Services -

“…Kim et al [97] proposed "The Power of Batching in the Click Modular Router." They investigated a method to run the Click in recent high-performance commodity servers consisting of multi-core CPUs in a NUMA architecture with multi-queue NIC.…”

“…Kim et al [97] (2012) Click on multi-core CPU. Batching in terms of both packet I/O and computation is crucial in a NUMA architecture with multi-queue NIC.…”

Fast and flexible: Parallel packet processing with GPUs and click

“…In contrast, Snap focuses on the routing performance of a single PC, and is thus complementary to RouteBricks; by increasing each PC's throughput, Snap could be used to reduce the number of PCs needed in a RouteBricks-like routing cluster. Kim et al demonstrated another strategy for improving Click performance by batching packets [13]. Here, batching refers to executing the same element on a set of packets in series, in contrast with standard Click, which executes a set of elements on the same packet in series.…”

“…The bottleneck in performance appears to be the large number of CPU elements in this configuration-there are 15 types of elements, some of which are duplicated sixteen times, once for each thread. As future work, we believe that the throughput can be significantly improved by moving some of these to the GPU and applying the techniques from Kim et al [13] to optimize the remaining CPU portions of the configuration.…”