By: Leo Zancani

Virtualized network functions are now seeing real deployments, delivering real services and thus graduating into the world of network operators’ mission critical systems.

Experience is showing that – even for these initial use-cases – NFV must not only coexist with the current infrastructure, services and operational activity, but also collaborate with them in some ways, for the fulfillment and assurance of services.

Alongside the inherent questions and challenges posed about the role of OSS in future networks by NFV (and its close sibling technology SDN), this makes the question of OSS readiness not just about OSS vendor roadmaps, but also about the condition of operational data and processes within CSP “legacy” networks.

Pattern of the rise of NFV

No one close to network operations is surprised to find that NFV is following the pattern of every other change that has taken hold in CSP infrastructure: evolution.

The precise form this evolution will take is as yet unclear; but looking at what was perhaps the last disruption on this scale to take place – IP networking – may yield some clues.

IP began to make a serious impact in the service provider space with the arrival of MPLS in the late 1990s; and yet, nearly twenty years later, complete displacement of traditional circuit-switched technology is only just now starting to look like a reality as technologies, such as packet over optical transport, become more widespread and intervening layers of legacy connectivity are eliminated.

Because the transition was inadequately planned in terms of the big picture view of provider operations, in the interim period, IP – despite its huge and indubitable value - added significant additional complexity to carrier networks and OSS, created yet another technology silo and caused a king’s ransom to be spent in integration and related operational costs.

Given the current state of play in trials and early deployments, it doesn’t seem unreasonable to expect that NFV – and indeed, SDN – risks following a similar pattern, albeit for different network layers.

In any case, a transitional period seems inevitable; and although there appear to be few voices that disagree with that view, there is precious little action. Standards bodies, operators and vendors alike are focused on demonstrating how the basic service lifecycle can be implemented using NFV and have not yet begun to seriously consider how it will grow out across their estates. This may well seem to be a reasonable stance given limited resources. After all, until some of the questions about what a practical, multi-vendor NFV platform looks like are answered, it is hard to properly imagine the challenges of managing it as part of operations overall.

However, as the story of IP illustrates, leaving these questions on hold is unwise, and can push huge unnecessary costs into many years of transitional operation, delaying (or even threatening to undermine altogether) the expected benefits of transformational technology.

The Role of OSS

In the infancy of NFV as an idea, much was made of the possibility that it might make OSS redundant. This is a bold enough statement to merit a closer look.

Definitions of OSS vary across industry bodies such as the TM Forum and the ITU and between different network operators, but it’s generally accepted to include all the systems which fall outside of the forwarding, control or management planes of the network itself, but which are nevertheless required to deliver and operate services and manage the network at scale.

In the future, or so the argument goes, much of this functionality will be dealt with locally and autonomously by the infrastructure itself.  Fault tolerance rather than fault management will become the norm and capacity will become elastic.

This may well eventually become the case; but as an overall view of how CSPs function, it overlooks two crucial facts.

Firstly, CSPs are still network operators: they still own infrastructure. It is possible that, eventually, the industry will reshape itself so that the business of owning and operating infrastructure will become entirely separate from that of delivering services; but, as things stand now – and for the foreseeable future – this remains in the arena of the CSPs. Only the purest of over-the-top players have the luxury of leaving that to ill-remunerated third parties. Virtualize all you like, but fibres will still be cut by construction workers and birds will still nest in the antennae of wireless links. The concerns of the physical infrastructure cannot be handled exclusively by the adaptations available to fault tolerant infrastructure; at some level, they must still be diagnosed and fixed by processes which are to some extent manual.