SLA-Aware Fine-Grained QoS Provisioning for Multi-Tenant Software-Defined Networks
Gaolei Li,Jun Wu,Jianhua Li,Zhenyu Zhou,Longhua Guo
DOI: https://doi.org/10.1109/access.2017.2761553
IF: 3.9
2017-01-01
IEEE Access
Abstract:Multi-tenant software-defined networks have emerged as a promising service platform to manage and share virtual network services between tenants and end-users. However, when using a service level agreement (SLA) in a multi-tenant circumstance, the resource competition among tenants incurred in many challenges on the quality of services (QoS) provisioning. First, as the existing QoS provisioning was to translate the high-level SLA contact into a group of corresponding low-level QoS rules, this traditional top-down translation mechanism was too ossified to support the dynamic SLA negotiation. Second, since an end-user usually requested multiple services and different services had different QoS requirements, the QoS provisioning required tenants to monitor and control the network status though OpenFlow protocol in a fine-grained way. To address these issues, we propose an SLA-aware fine-grained QoS provisioning (SFQP) scheme for MTSDN. The SFQP scheme extracts the eigen characteristics of each packet by application-aware technology automatically. Considering tenants are more willing to serve for users which have popular SLA contact to earn more revenues, we predict the SLA popularity by K-nearest neighbor. Moreover, we present an SLA-aware dynamic bandwidth allocation algorithm to implement the proposed SFQP scheme. The feasibility of the proposed scheme is verified by simulations of the SLA taxonomy, bandwidth utilization, and bandwidth fluctuation.