Analysis of Latency on Network Connection Mediums Using A Public Cloud-Based Transactional Database

Abstract

Computing applications that utilize public cloud services are known to have latency (delay) caused by physical distances between clients and public cloud providers, coupled with delays created by intermediate routers, and the physical connection medium itself. Latencies caused by the physical medium used to connect to these cloud services have been vastly untested. For instance, a business that uses cloud services through a cable Internet connection might experience acceptable latencies, while a business using DSL might experience unacceptable delays. This study uses quantitative research and empirical testing to determine how various connection mediums have different performance characteristics that affect latency, as well as reveals how slower connection mediums may not be able to support cloud services due to the latency accumulated through the use of a cloud-based transactional database. These tests will provide a means to determine how varying the connection medium to a cloud service provider will affect download and upload latency to a public cloud database. Three connection mediums are being tested for latency; cable Internet, DSL, and a fiber optic backbone connection. This research will serve as a tool for businesses interested in DSL, cable, and fiber optic Internet connection mediums to help determine the amount of latency they will experience by using a public cloud transactional database (hosted by Microsoft Azure).