Virtual Resource Description and Clustering for Virtual Network Discovery

Abstract

Virtual resource description and clustering for virtual network discovery.

Adaptive virtual network provisioning

“…Hence, VN extensibility and elasticity are critical, especially in dynamically changing network environments. Existing work on network virtualization mainly focuses on initial VN provisioning including resource matching [19,15], VN embedding [25,24,19,18,10,14] and VN binding [7,21,13,9,20]. According to our knowledge, there are no research studies on adaptive provisioning of instantiated VNs to cope with dynamic changes in service demands and resource availability, especially when these changes enforce dynamic virtual node and link re-assignment.…”

“…2 -Step 1) to inform VN Providers on available substrate resources properties. In previous work [15], a resource description schema enables InPs to describe and customise their offered substrate resources. Physical resources are described in terms of functional attributes (i.e., static parameters such as node/link type) and non-functional attributes (corresponding to real-time parameters such as currently available capacity/bandwidth).…”

“…2, VN provisioning includes four steps [15]: (i) resource description and advertisement, (ii) resource discovery and matching, (iii) VN embedding, and (iv) VN binding. Hereby, we briefly discuss these steps, since initial provisioning is extended to achieve adaptive embedding and handle run-time dynamic variations.…”

“…2 (Steps 4, 5), the VN Provider uses the discovery framework to search, discover and match the available substrate resources advertised by the InPs with the VN requests. To facilitate the overall discovery and matching process, in [15] we organized and classified the resource descriptions registered in the discovery framework using conceptual clustering. This groups substrate resources with similar concepts, descriptions and properties into clusters.…”

“…Such clustering generates a hierarchy of clusters, called dendrogram, along with conceptual descriptions of each cluster. A similarity based matching algorithm [15] associates the required virtual nodes with the most similar cluster in the dendrogram. As a result, the initial matching step identifies for each required VN node (in the request) candidate resources (in the substrate) that fulfil the virtual node requirements.…”