Analysis of Scaling Policies for NFV Providing 5G/6G Reliability Levels With Fallible Servers

Abstract

The softwarization of mobile networks enables an efficient use of resources, by dynamically scaling and re-assigning them following variations in demand. Given that the activation of additional servers is not immediate, scaling up resources should anticipate traffic demands to prevent service disruption. At the same time, the activation of more servers than strictly necessary results in a waste of resources, and thus should be avoided. Given the stringent reliability requirements of 5G applications (up to 6 nines) and the fallible nature of servers, finding the right trade-off between efficiency and service disruption is particularly critical.In this paper, we analyze a generic auto-scaling mechanism for communication services, used to de(activate) servers in a cluster, based on occupation thresholds. We model the impact of the activation delay and the finite lifetime of the servers on performance, in terms of power consumption and failure probability. Based on this model, we derive an algorithm to optimally configure the thresholds. Simulation results confirm the accuracy of the model both under synthetic and realistic traffic patterns as well as the effectiveness of the configuration algorithm. We also provide some insights on the best strategy to support an energy-efficient highly-reliable service: deploying a few powerful and reliable machines versus deploying many machines, but less powerful and reliable.

Energy-Aware Adaptive Scaling of Server Farms for NFV With Reliability Requirements

“…target failure probabilities T f = {10 −3 , 10 −5 } and evaluate the resulting failure probability and power consumption provided by A3S for the 'rack servers' configuration. As a benchmark, we consider the static optimal configuration provided by our previous work [25], which assumes Poisson arrivals and exponential service times. We summarize the obtained results in Table 2.…”

Enabling Dynamic Autoscaling for NFV in a Non-Standalone Virtual EPC: Design and Analysis

“…In the literature, various approaches have been proposed to predict the VM load to boot VMs before these VMs, which are under operation, become overloaded. These approaches can be categorized into those based on the moving average, including the exponential weighted moving average (EMA) [8], [14], autoregressive moving average (ARMA) [15], [16], and autoregressive integrated moving average (ARIMA) [17], [18], those based on machine learning [19], those based on Markov models [20], [21], and those based on queueing models [22]- [30]. In the following, we summarize each category briefly.…”

“…Queueing model-based approaches were studied in [22]- [30]. In their developed models, the authors considered the setup time both with [22] and without [23] defections.…”

“…These mechanisms either ignore the VM setup time or deal only with virtualized resources without considering resources from legacy systems. Although [27], [28], [30] derived optimal properties and presented a precise analysis for a queueing model, [28] ignores setup time and [27], [30] did not consider legacy servers. Therefore, the existing mechanisms and models are not applicable to non-standalone 5G networks, where eNBs and gNBs coexist.…”

Edge Computing Platform with Efficient Migration Scheme for 5G/6G Networks

“…System components place considerable limits on context, information, and mobility to ensure the best system accessibility and reliability and the ultra-low latency required for haptic Internet systems [13,14]. Industry and research organizations are constantly seeking to create and apply novel technologies and models in communication networks, such as the distributed edge computing, e.g., mobile edge computing (MEC) and fog computing, artificial intelligence (AI), software-defined networking (SDN), blockchain, and network function virtualization (NFV), in order to maintain growth and fulfill ever-changing demands [15][16][17][18].…”