Hierarchical Multiresource Fair Queueing for Packet Processing

Abstract

Various middleboxes are ubiquitously deployed in networks to perform packet\nprocessing functions, such as firewalling, proxy, scheduling, etc., for the\nflows passing through them. With the explosion of network traffic and the\ndemand for multiple types of network resources, it has never been more\nchallenging for a middlebox to provide Quality-of-Service (QoS) guarantees to\ngrouped flows. Unfortunately, all currently existing fair queueing algorithms\nfail in supporting hierarchical scheduling, which is necessary to provide QoS\nguarantee to the grouped flows of multiple service classes. In this paper, we\npresent two new multi-resource fair queueing algorithms to support hierarchical\nscheduling, collapsed Hierarchical Dominant Resource Fair Queueing (collapsed\nH-DRFQ) and dove-tailing H-DRFQ. Particularly, collapsed H-DRFQ transforms the\nhierarchy of grouped flows into a flat structure for flat scheduling while\ndove-tailing H-DRFQ iteratively performs flat scheduling to sibling nodes on\nthe original hierarchy. Through rigorous theoretical analysis, we find that\nboth algorithms can provide hierarchical share guarantees to individual flows,\nwhile the upper bound of packet delay in dove-tailing H-DRFQ is smaller than\nthat of collapsed H-DRFQ. We implement the proposed algorithms on Click modular\nrouter and the experimental results verify our analytical results.\n