Future 5G networks will run into the massive spread of applications with diverse and stringent requirements in terms of latency, reliability and security. Current multi-layer transport networks lack coordination between their constituent IP/MPLS and optical layers, and thus cannot provide to applications a full service differentiation all the way down to the optical layer. This limitation can be overcome by the adoption of a Software-Defined Network (SDN) orchestrator between the underlying multi-layer transport network and the applications, with the aim of offering them tailored network connectivity and minimizing network resources' consumption. The orchestrator must thus be able to dynamically compute multi-layer paths and allocate optical and IP/MPLS resources in an application-aware fashion (i.e, to meet applications' requirements), both when applications require service provisioning for the first time and when their connectivity must be restored after any network failure. In this paper, we propose a novel auxiliary-graph-based application-aware service provisioning algorithm, able to dynamically provision applications' service requests at runtime according to their needs. Simulations show that, unlike a benchmark application-unaware algorithm, it prevents the violation of application requirements (in terms of bandwidth, latency and availability) while guaranteeing a similar blocking probability. Following the same principles, we then propose an application-aware restoration algorithm, where fine-grained application traffic flows can be individually restored according to their tolerance to service disruption: a faster IP/MPLS restoration is considered more appropriate for applications with low tolerance, while slower multi-layer restoration is chosen for the others. Also this approach is proved to always comply with application requirements, and it can successfully restore much more application traffic flows than simple optical restoration.

Application-aware Service Provisioning and Restoration in SDN-based Multi-layer Transport Networks

MArco Savi;siracusa domenico
2018

Abstract

Future 5G networks will run into the massive spread of applications with diverse and stringent requirements in terms of latency, reliability and security. Current multi-layer transport networks lack coordination between their constituent IP/MPLS and optical layers, and thus cannot provide to applications a full service differentiation all the way down to the optical layer. This limitation can be overcome by the adoption of a Software-Defined Network (SDN) orchestrator between the underlying multi-layer transport network and the applications, with the aim of offering them tailored network connectivity and minimizing network resources' consumption. The orchestrator must thus be able to dynamically compute multi-layer paths and allocate optical and IP/MPLS resources in an application-aware fashion (i.e, to meet applications' requirements), both when applications require service provisioning for the first time and when their connectivity must be restored after any network failure. In this paper, we propose a novel auxiliary-graph-based application-aware service provisioning algorithm, able to dynamically provision applications' service requests at runtime according to their needs. Simulations show that, unlike a benchmark application-unaware algorithm, it prevents the violation of application requirements (in terms of bandwidth, latency and availability) while guaranteeing a similar blocking probability. Following the same principles, we then propose an application-aware restoration algorithm, where fine-grained application traffic flows can be individually restored according to their tolerance to service disruption: a faster IP/MPLS restoration is considered more appropriate for applications with low tolerance, while slower multi-layer restoration is chosen for the others. Also this approach is proved to always comply with application requirements, and it can successfully restore much more application traffic flows than simple optical restoration.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11582/314803
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
social impact