Network Function Virtualization (NFV) enables operators/carriers to deploy and execute new network node services as virtualized software and eliminates the need for dedicated and possibly under-utilized hardware. In this architecture, networking functions are virtualized (VNFs) and chained together to provide an end-to-end network service. OpenStack as a Virtual Infrastructure Manager is one of the key enablers for realizing the NFV Management and Network Orchestration architecture.
Some of the key challenges in NFV Orchestration for service chains are
- Scheduling of a service chain of VNFs across several Compute Nodes or OpenStack powered clouds. This is required to determine where the different network functions of a service chain are to be located
- Determining an optimal method to scale the network functions if a virtualized element (or a group of elements) of the service chain is shared across multiple users or clients
- Any automated network optimization or resource management would require optimization of the entire network, including the legacy or non-virtualized component of the end-to-end service. Currently, the orchestrators are not always designed to handle this.
- The orchestrator, which is generic, needs to implement certain standardized interfaces towards VNF and OSS/BSS to provide automated solutions.
In this paper, we evaluate possible models and solutions to determine efficient ways to perform resource management and placement of service chains. The paper also summarizes the recent research in this area and future direction for the OpenStack community interested in efficient NFV deployments using OpenStack.