Semantic validation of affinity constrained service function chain requests

Abstract

Abstract-Network Function Virtualization (NFV) has been proposed as a paradigm to increase the cost-efficiency, flexibility and innovation in network service provisioning. By leveraging IT virtualization techniques in combination with programmable networks, NFV is able to decouple network functionality from the physical devices on which they are deployed. This opens up new business opportunities for both Infrastructure Providers (InPs) as well as Service Providers (SPs), where the SP can request to deploy a chain of Virtual Network Functions (VNFs) on top of which its service can run. However, current NFV approaches lack the possibility for SPs to define location requirements and constraints on the mapping of virtual functions and paths onto physical hosts and links. Nevertheless, many scenarios can be envisioned in which the SP would like to attach placement constraints for efficiency, resilience, legislative, privacy and economic reasons. Therefore, we propose a set of affinity and anti-affinity constraints, which can be used by SPs to define such placement restrictions. This newfound ability to add constraints to Service Function Chain (SFC) requests also introduces an additional risk that SFCs with conflicting constraints are requested or automatically generated. Therefore, a framework is proposed that allows the InP to check the validity of a set of constraints and provide feedback to the SP. To achieve this, the SFC request and relevant information on the physical topology are modeled as an ontology of which the consistency can be checked using a semantic reasoner. Enabling semantic validation of SFC requests, eliminates inconsistent SFCs requests from being transferred to the embedding algorithm.

Semantically Enhanced Mapping Algorithm for Affinity-Constrained Service Function Chain Requests

“…None of the aforementioned approaches offers support for attaching affinity or anti-affinity constraints to the SFCs nor do they take into account such constraints when evaluating the embeddings. This paper is an extension of previous work [35] in which the affinity and anti-affinity constraints were first proposed. The previous paper proposed a first version of the semantic framework for affinity-constrained SFCs validation.…”

“…In previous work we proposed to exploit ontology representations for the purpose of modeling the physical substrate, the SFC requests and defining a set of rules that can be used to infer additional information [35]. Figure 3 represents the proposed semantic model.…”

“…If a VirtualEdge z contains a VNF x embedded at Location a, this VNF z is embedded at Location a as well (Rule (3)). For a full description of the rules, we refer to previous work [35] which focuses on the semantic validation of SFC requests. …”

Optimal orchestration of virtual network functions

“…• VNF affinity and anti-affinity rules: due to the privacy, reliability or other reasons, a provider may want impose rules on the placement of certain types of VNF: be placed or not placed on certain servers, be grouped or not grouped together, etc. Such specific VNF placement rules are called affinity and/or anti-affinity rules [35]. To extend our model to take them into account, the simplest way is to introduce a new variable representing the presence of a certain type of VNF f on a given node i (we remind the reader that our model allows to have multiple copies of the same type of VNF on the same node).…”

ReasoNet: Inferring Network Policies Using Ontologies

“…Neo4j provides excellent support for graph processing, but it lacks reasoning and consistency validation automation capabilities. [31] uses semantic web technologies to model physical substrate resources, service function chain and deployment constraints for virtual function embedding.…”