The Fork-Join Terabit per Second Router

Internet同时面临着两个问题：我们需要更快的交换/路由基础设施，以及我们需要引入可扩展的服务质量（QoS）。作为一个社区，我们对每个建议：我们可以使路由器更快地使用输入排队交叉开关，而不是共享内存系统，我们可以引入QoS使用基于WESTERN（加权公平交换）的数据包调度。但我们不知道如何同时做到这两点。到目前为止，这两种解决方案一直是相互排斥的-所有基于WESTERN的调度算法的工作都要求交换机/路由器使用输出共享或集中式共享内存。该路由器是一个非常高性能、可扩展、容错的系统，旨在以每秒多兆比特的速度运行，同时提供对服务质量的精确支持。该路由器是高度并行的，并且由多个相同的独立操作的平面组成。当它们到达时，属于同一个流的所有数据包都被定向（分叉）到路由器的同一个平面进行处理。在fork-join路由器能够被构建（作为原型，或者在工业上作为商业产品）之前，必须回答一些重要的问题：如何实现fork、join和switch调度功能？我们的工作重点是回答这些问题。

The Internet is facing two problems simultaneously: we need a fasterswitching/routing infrastructure, and we need to introduce guaranteedqualities of service (QoS). As a community, we have proposals for each: wecan make the routers faster by using input-queued crossbar switches,instead of shared memory systems; and we can introduce QoS using WFQ-based(weighted fair queueing) packet scheduling. But we don't know how to doboth at the same time. Until now, the two solutions have been mutuallyexclusive - all of the work on WFQ-based scheduling algorithms has requiredthat switches/routers use output-queueing, or centralized shared memory.The Fork-Join Router is designed to address both problems. The router is avery high performance, scalable, fault-tolerant system designed to operateat multi-terabits per second, yet provide exact support for qualities ofservice. The router is highly parallel, and is composed of multiple,identical, independently operating planes. When they arrive, all of thepackets belonging to the same flow are directed (forked) to the same planeof the router for processing. Prior to departure, the flows are remerged(joined) soas to maintain the QoS of each flow through the system.Before a fork-join router can be built (as a prototype, or in industry as acommercial product) some significant questions must be answered: How canthe fork, join and switch scheduling functions be implemented? Our workfocusses on answering these problems.