Performance Modeling of Softwarized Network Services Based on Queuing Theory With Experimental Validation

Network Functions Virtualization facilitates the automation of scaling softwarized network services (SNSs). However, realization such a scenario requires way to determine needed amount resources so that SNSs performance requisites are met for given workload. This problem is known as resource dimensioning, and it can be efficiently tackled by modeling. In this vein, paper describes an analytical model based on open queuing G/G/m queues evaluate response time SNSs. We validate our experimentally virtualized Mobility Management Entity (vMME) with three-tiered architecture running testbed resembles typical data center virtualization environment. detail description experimental setup procedures. solve resulting queueing using Queueing Networks Analyzer (QNA), Jackson's networks, Mean Value Analysis methodologies, compare them in terms estimation error. Results show that, medium high workloads, QNA method achieves less than half error compared standard techniques. For low three methods produce lower 10 percent. Finally, we usefulness performing dynamic provisioning vMME experimentally.