The Analysis of Traffic Balancing for Data Centers Serving Requests of LEO Mobile Satellite Systems

One of the features of low earth orbit (LEO) mobile satellite systems is to deliver information to the data analytic centers. Many types of alarm information are collected by smart meters and video cameras that are placed where wired telecommunication is not possible due to the technical or economic meanings. It is supposed that requests for servicing are arriving to the data center following the Poisson model and each requiring exponentially distributed computing time. A data center consists of some number of computing units. Each unit processes the requests according to the Processor Sharing (PS) service discipline and cannot accept at the same time more than fixed number of requests. The mathematical model of load balancing between computing units is constructed. Two scenarios are considered: “static” when requests are routed to computing units according to the prescribed probabilities and “greedy” when requests are routed to the least loaded computing unit. Using the model, the main performance measures of interest are given with help of values of probabilities of model's stationary states. The recursive algorithm of performance measures estimation is suggested. The model and derived algorithms can be used for study the scenarios of traffic balancing at the data analytic centers.