A Study on Robustness for the Single Server Location in Distributed Interactive Applications

In distributed interactive application ecosystems, service provision and availability are key factors for the end user's perceived quality. Therefore, the location of any server facility is important from the provider's standpoint for efficient capital expenditure. Given the dynamicity of such environments (e.g., due to traffic load variance over time), modeling this challenge as a median problem may require the continuous calculation of the optimal location. To address this robustness issue, a new robustness metric is introduced here for the case of the single server in edge computing, which depends on the overall delay experienced by the data traffic. Moreover, the smallest eigenvalue of the median matrix is considered as an alternative means for measuring this robustness. Using trace-driven simulation results, it is demonstrated that there is indeed a relationship between the proposed robustness metric and the smallest eigenvalue for the under consideration environment.