We present P5; an event-driven policy framework that allows network operators to realize end-to-end policies on top of P4-based data planes in an intuitive and effective manner. We demonstrate how P5 adheres to a service-level agreement (SLA) by applying P4-based traffic engineering with latency constraints.

Deployment of 100Gigabit Ethernet (GbE) links challenges the packet processing limits of commodity hardware used for Network Functions Virtualization (NFV). Moreover, realizing chained network functions (i.e., service chains) necessitates the use of multiple CPU cores, or even multiple servers, to process packets from such high speed links. Our system Metron jointly exploits the underlying network and commodity servers’ resources: ( i ) to offload part of the packet processing logic to the network, ( ii ) by using smart tagging to setup and exploit the affinity of traffic classes, and ( iii ) by using tag-based hardware dispatching to carry out the remaining packet processing at the speed of the servers’ cores, with zero inter-core communication. Moreover, Metron transparently integrates, manages, and load balances proprietary “blackboxes” together with Metron service chains. Metron realizes stateful network functions at the speed of 100GbE network cards on a single server, while elastically and rapidly adapting to changing workload volumes. Our experiments demonstrate that Metron service chains can coexist with heterogeneous blackboxes, while still leveraging Metron’s accurate dispatching and load balancing. In summary, Metron has ( i ) 2.75–8× better efficiency, up to ( ii ) 4.7× lower latency, and ( iii ) 7.8× higher throughput than OpenBox, a state-of-the-art NFV system.

While the current literature typically focuses on load-balancing among multiple servers, in this paper, we demonstrate the importance of load-balancing within a single machine (potentially with hundreds of CPU cores). In this context, we propose a new load-balancing technique (RSS++) that dynamically modifies the receive side scaling (RSS) indirection table to spread the load across the CPU cores in a more optimal way. RSS++ incurs up to 14x lower 95th percentile tail latency and orders of magnitude fewer packet drops compared to RSS under high CPU utilization. RSS++ allows higher CPU utilization and dynamic scaling of the number of allocated CPU cores to accommodate the input load while avoiding the typical 25% over-provisioning.

Recent approaches to network functions virtualization (NFV) have shown that commodity network stacks and drivers struggle to keep up with increasing hardware speed. Despite this, popular cloud networking services still rely on commodity operating systems (OSs) and device drivers. Taking into account the hardware underlying of commodity servers, we built an NFV profiler that tracks the movement of packets across the system’s memory hierarchy by collecting key hardware and OS-level performance counters. Leveraging the profiler’s data, our Service Chain Coordinator’s (SCC) run-time accelerates user-space NFV service chains, based on commodity drivers. To do so, SCC combines multiplexing of system calls with scheduling strategies, taking time, priority, and processing load into account. By granting longer time quanta to chained network functions (NFs), combined with I/O multiplexing, SCC reduces unnecessary scheduling and I/O overheads, resulting in three-fold latency reduction due to cache and main memory utilization improvements. More importantly, SCC reduces the latency variance of NFV service chains by up to 40x compared to standard FastClick chains by making the average case for an NFV chain to perform as well as the best case. These improvements are possible because of our profiler’s accuracy.

Autonomic network management systems will operate in a volatile network environment; thus they should be able to continuously adapt their decision making mechanism through learning from the behavior of the communication system. In this paper, a novel learning scheme is proposed based on the network-wide collected performance experience, targeting the enhancement of network elements' decision making engine. The algorithm employs a fuzzy logic inference engine in order to enable self-managed network elements faults or optimization opportunities identification, which is enhanced by applying data mining techniques on the accumulated observations.