e-Infrastructure and e-Services for Developing Countries. 10th EAI International Conference, AFRICOMM 2018, Dakar, Senegal, November 29-30, 2019, Proceedings

Research Article

Performance Barriers to Cloud Services in Africa’s Public Sector: A Latency Perspective

  • @INPROCEEDINGS{10.1007/978-3-030-16042-5_15,
        author={Josiah Chavula and Amreesh Phokeer and Enrico Calandro},
        title={Performance Barriers to Cloud Services in Africa’s Public Sector: A Latency Perspective},
        proceedings={e-Infrastructure and e-Services for Developing Countries. 10th EAI International Conference, AFRICOMM 2018, Dakar, Senegal, November 29-30, 2019, Proceedings},
        proceedings_a={AFRICOMM},
        year={2019},
        month={3},
        keywords={Latency Internet measurements Cloud services},
        doi={10.1007/978-3-030-16042-5_15}
    }
    
  • Josiah Chavula
    Amreesh Phokeer
    Enrico Calandro
    Year: 2019
    Performance Barriers to Cloud Services in Africa’s Public Sector: A Latency Perspective
    AFRICOMM
    Springer
    DOI: 10.1007/978-3-030-16042-5_15
Josiah Chavula1,*, Amreesh Phokeer2,*, Enrico Calandro3,*
  • 1: University of Cape Town
  • 2: AFRINIC
  • 3: Research ICT Africa
*Contact email: jchavula@cs.uct.ac.za, amreesh@afrinic.net, ecalandro@researchictafrica.net

Abstract

Cloud computing allows individuals and organizations to lease storage and computation resources remotely and as needed. For such remote access to work efficiently, there is need for reliable and low-delay delivery of Internet traffic. By carrying out month-long Internet measurement campaign, this paper investigates location and latencies of cloud-based web hosting in the public sector of five African countries. Results of the study show that a large percentage of public sector websites are hosted in cloud-based infrastructure physically located in America and Europe. Analysis of latencies shows significant differences between local and remote hosted websites, and that delays are significantly lower for countries that host CDN nodes. The results also indicate high delays for local websites that are accessed circuitously.