Model-based analytics for profiling workloads in virtual network functions

Abstract

With the flexibility and programmability levels offered by Network Functions Virtualization (NFV), it is expected to catalyze the upcoming "softwarization" of the network through software implementation of networking functionalities on virtual machines (VMs). While looking into the different issues thrown at NFV, numerous works have demonstrated how performance, power consumption and, consequently, the optimal resource configuration and VM allocation vary with the statistical features of the workload-specifically, the "burstiness" of the traffic. This paper proposes a model-based analytics approach for profiling (virtual) network function (VNF) workloads that captures traffic burstiness, considering-and adding value to-hardware/software performance monitor counters (PMCs) available in Linux host servers. Results show good estimation accuracies for the chosen PMCs, which can be useful to enhance current methods for finegrained provisioning, usage-based pricing and anomaly detection, and facilitate the way towards an agile network.

A Model-Based Approach Towards Real-Time Analytics in NFV Infrastructures

“…The energy-aware core hosting the VNF (or VNFC) is modeled as an M X /G/1/SET queue, as in [6] [8] [9]. This model generalizes the well-known M X /G/1 queue [33] for burst arrivals, by also covering the cases in which an additional setup period SET is necessary before service can be resumed.…”

“…Linux has different utilities for performance monitoring -among them, the PMCs described in the following are considered in this work; other PMCs used in [6] did not work well with the BMAP emulation. Note that in the syntax of these utilities, the term 'CPU' refers to a core (or logical core, in the case of hyperthreading).…”

“…Rather than simply tuning deterministic parameters in the Ixia NX2 traffic generator, as in [6], incoming traffic is more accurately generated to emulate BMAP arrivals in this extended version. Tcl scripts are used to specify batch inter-arrival times and batch sizes.…”

“…As initially motivated in [6], the proposed approach seeks to profile VNF workloads beyond offered loads and utilization -specifically, to capture the workload burstiness, as characterized by the model parameters λ, β (1) and β (2) .…”

“…An initial version of this work has been presented in [6], in which various PMCs are evaluated for the black-box estimation of key statistical features of the VNF workload, considering a fairly general renewal model (M X /G/1/SET queue [7]) that captures traffic burstiness and system setup times. In this extended version, we provide a complete analytical characterization of the M X /G/1/SET queue, which includes power and latency models.…”