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

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