Network Function Offloading Through Classification of Elephant Flows

Abstract

With the move from traditional hardware appliance based network functions to Network Function Virtualization, software development is decoupled from the hardware. However, as a network function is no longer optimized for hardware, beneficial features of networking hardware may not be used any more. Solutions such as SDN or NIC offloading aim to overcome this antipodes by integrating networking hardware into the packet processing pipelines. On the one hand, offloading traffic of network functions to hardware can increase throughput and reduce resource consumption. On the other hand, the number of parallel flows in a network can be very high, exhausting the capacity of the tables of the networking hardware and force the system to fall back to software processing. Fortunately, it is known that a large portion of the flows in the internet are mice flows, whereas the majority of the traffic is constituted by elephant flows. If the elephant flows can be detected efficiently, the hardware tables can be used more efficiently as a larger share of the traffic can be offloaded. We introduce a machine learning based approach that takes its decision with the first packet of a flow. A fundamentally different approach is using packet sampling for the offloading decision. We are evaluating both approaches in terms of complexity, offloaded share of the traffic and table occupation. The results show that a machine learning based offloading decision is possible with the first packet. The sampling approach only reaches a comparable performance at very high sampling rates.

ComPipe: A Novel Flow Placement and Measurement Algorithm for Programmable Composite Pipelines

“…CacheFlow [17] constructs a directed acyclic graph model to identify and analyze dependencies between rules, employing either the Dependent-Set Algorithm or the Cover-Set Algorithm to determine which rules should be cached in hardware. LFP [18] proposes a method for network function offloading, blending sampling and machine learning to decide on hardware node offloading based on initial flow packet predictions. OVS-CAB [7] presents an offloading mechanism for Open vSwitch on smart network interface cards, addressing rule overlap issues effectively.…”

Elephant Flow Classification on the First Packet With Neural Networks

“…In contrast, our primary objective is to identify a flow as rapidly as possible, ideally from its first packet. Durner et al in [20] demonstrated flow classification on the first packet, using only features extracted from the 5-tuple (src IP, dst IP, src port, dst port, protocol) and the size of the first packet. In [21] Hardegen et al suggested multiclass prediction as an alternative to binary classification (elephant/mouse), employing a deep neural network to predict flow characteristics from the 5-tuple of the first packet.…”

HyperNAT: Scaling Up Network AddressTranslation with SmartNICs for Clouds

“…Durner et. al. proposes to distribute traffic between NICs and the host according to flow characteristics [6]. Flex-Gate [19] proposes to offload proper functionalities among diverse ones to the NIC and the remaining are on the end host.…”