Providing Full QoS Support in Clusters Using Only Two VCs at the Switches

Abstract

Abstract. Current interconnect standards providing hardware support for quality of service (QoS) consider up to 16 virtual channels (VCs) for this purpose. However, most implementations do not offer so many VCs because they increase the complexity of the switch and the scheduling delays. In this paper, we show that this number of VCs can be significantly reduced. Some of the scheduling decisions made at network interfaces can be easily reused at switches without significantly altering the global behavior. Specifically, we show that it is enough to use two VCs for QoS purposes at each switch port, thereby simplifying the design and reducing its cost.

Scalable low-cost QoS support for single-chip switches

“…We will also give details on the parameters of the network and the load used for the evaluation. The main differences with the study at [11] are:…”

“…In [11] we have proposed a new strategy to use only two VCs at each switch port to provide QoS. It achieves similar performance results to those using many more VCs.…”

“…• In [11] we considered a theoretical model for scheduling times, while here we take an approach based on the actual design of the switch.…”

“…In [11], we propose a strategy to use just two VCs at each switch port for the provision of QoS that allows very similar results as if we were using many more VCs. One of these two VCs is used for QoS traffic and the other one for best-effort traffic.…”

A Hardware NIC Scheduler to Guarantee QoS on High Performance Servers

“…A way to reduce router complexity and to optimize its design, while meeting QoS constraints, consists in improving traffic scheduling in servers [10]. The main idea is to reduce the number of buffers and VCs, taking advantage of QoS traffic sorting carried out by servers.…”