Characterization and analysis of cloud-to-user latency: The case of Azure and AWS

Abstract With the growing adoption of cloud infrastructures to deliver a variety of IT services, monitoring cloud network performance has become crucial. However, cloud providers only disclose qualitative information about network performance, at most. This hinders efficient cloud adoption, resulting in uncertainties about the behavior of hosted services, and sub-optimal deployment choices. In this work, we focus on cloud-to-user latency, i.e. the latency of network paths interconnecting datacenters to worldwide-spread cloud users accessing their services. Specifically, we performed a 14-day measurement campaign from 25 vantage points deployed via the Planetlab infrastructure (emulating spatially-spread users) and considering services running in distinct locations on the infrastructures of Amazon Web Services and Microsoft Azure. First, our experimentation allows us to provide an in-depth performance characterization (based on multiple probing methods and fine-grained sampling rate) of such networks as perceived by users spread worldwide, highlighting both spatial and temporal latency trends. Then, our analysis is exploited with design purposes to support both cloud customers and providers with the assessment of cloud-network performance (via badness detection & imputation tools) and the making of deployment decisions (via the evaluation of multi-cloud benefits). The dataset gathered from the campaign is publicly released to foster reproducibility.