The Nearest Origin-Shield (NOS): A Jitter-Free Overlay Routing Framework for Content Delivery Networks


Najaflou N., Aris A., Canberk B., Aydin Z. G.

International Symposium on Networks, Computers and Communications (ISNCC), İstanbul, Türkiye, 18 - 20 Haziran 2019 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Cilt numarası:
  • Doi Numarası: 10.1109/isncc.2019.8909164
  • Basıldığı Şehir: İstanbul
  • Basıldığı Ülke: Türkiye
  • İstanbul Teknik Üniversitesi Adresli: Evet

Özet

Although Content Delivery Networks (CDN) do their best to quickly deliver the contents, Internet Service Providers (ISP) and network conditions cause unexpected changes on the routing paths. Hence, it gets difficult for CDNs to adhere to the delay promises within the Service Level Agreements. This paper presents a jitter-free overlay routing framework for CDNs, which struggle to adapt timely content delivery to end users due to the dynamicity of the Internet. The Nearest Origin-Shield (NOS) framework which we propose in this paper overcomes this issue by creating an overlay network on top of the existing physical ISP networks and dynamically determining the fastest routes to the content sources (Origins) on this overlay network. NOS keeps track of changes in underlay network through periodically measuring the end-to-end routing paths delays via Helper Modules. Based on the underlay network delay measures and the load (i.e., CPU, disk I/O and network usage) of the overlay network entities, NOS Central Module determines the fastest routes to the Origins using our novel Delay-Aware and Jitter-Free Overlay Routing Algorithm. Hence making it possible for CDNs to provide jitter-free and timely content deliveries meanwhile ensuring the cache servers not to be overloaded. We created a new testbed for performance evaluations with real CDN servers, Origin accounts and measurements. Results show that, NOS provides shorter, more stable and jitter-free routing paths with jitter gains up to 98% and improves the Hit ratio by 0:74%