Negotiating on-demand connectivity between clouds and Wide Area Networks

Infrastructure as a Service (IaaS) provides the capability to deploy infrastructure on demand, but today there are gaps in the IaaS model for network connectivity, especially when user and/or application require guaranteed connectivity between the deployed infrastructures. This problem is aggravated when virtual infrastructures are geographically distributed, for performance and redundancy reasons. Connectivity is also crucial to acquire and integrate virtual resources in remote cloud datacenters with on-site IT resources. In enterprise environments where connectivity requirements cannot be met by best effort networks like Internet, such distributed infrastructure needs to make use of VPN services for inter-domain connectivity, provided by service providers over their trusted networks. To establish this connectivity, information regarding the interconnection link, such as interfaces and underlying protocols need to be negotiated and agreed upon. Today there is a lack of a technology independent mechanism and interface to perform such negotiation. This is a hindrance to automation of on-demand connectivity. In this paper we present a protocol for dynamic negotiation of connectivity service between domains. We describe how it has been used for negotiations between OpenStack and OpenNebula cloud datacenters and MPLS-based WAN, for establishment of L2/L3 VPN services.