According to Wikipedia, code refactoring “is the process of restructuring existing computer code – changing the factoring – without changing its external behavior.” Refactoring is the key to success with network function virtualization (NFV).
The industry is focused on virtualizing network functions, however virtualization itself won’t change much. Simply moving functions off a monolithic network element from within its current technology and organization boundaries and into software provided by a different vendor running on a virtual machine will likely increase costs and complexity – as the leading vendors of monolithic equipment are keen to point out.
The desired simplification and operational efficiencies can be realized by examining the network functions required and the current functional boundaries, then refactoring the service delivery components in an optimal fashion. As networks have evolved over time, functional boundaries between applications, operational support systems and network infrastructure have resulted in duplication of functions, complex protocols and specialized components to manage interaction across these boundaries. In service provider networks, there are many function- and technology-specific protocols that have been developed for specific types of infrastructures requiring specialized skills to configure and manage.
Refactoring entails reorganizing network functions and moving the resulting software components onto general-purpose NFV infrastructure, while carefully scrutinizing the need to maintain existing protocols. Done properly, refactoring allows operators to achieve the key NFV objectives of reducing both complexity and cost. In data centers, web services technology has democratized the protocols and mechanisms used for communications between devices and subsystems, resulting in rapid development and deployment of dynamically orchestrated cloud infrastructures.
This is where a network refactoring solution comes into play. By scrutinizing the components that are used to orchestrate dynamic broadband service delivery at the edge of the network, such a solution eliminates duplicate functions, removes complex application-specific technology, simplifies operations and provides developer-friendly web services APIs.
The cost reductions motivating operators to implement NFV cannot be achieved simply by deploying general-purpose platforms instead of monolithic network elements based on custom silicon. The only way to realize the promise of NFV is to first refactor the network, reducing the number of software components – yielding a corresponding reduction in complexity and cost.
So on the road to NFV, it’s worth asking: are existing vendors organizationally structured and motivated to fully embrace network refactoring?
This article was originally published in Active Broadband Network's Blog.
