Everyone knows that SDN and NFV are supposed to reduce cost, and many know that there's been a debate on whether capex or opex will win out. Right now, opex seems to be winning, but as it turns out, there are a lot of dimensions to opex and it's important to sort out the best costs to target before making a business case.
Finding the target
Most vendors see justifying SDN or NFV through opex reductions as boiling the ocean, but it doesn't have to be. It's true that opex distribution has a lot of categories. There are three broad classes of network operator, such as mobile, wireline and virtual (MSP and MVNO); and there are six classifications of what might be called "process opex," the opex that's related to operator service and network processes and not to things like paying roaming charges. In addition, vendors could also target big total cost, or where big cost changes are happening.
These multiple opex categories and distributions explain why some operators are telling glowing stories of software-defined networking (SDN) or network functions virtualization (NFV) success while others say they can't yet make a business case. Some operators are able to target an important cost component with a fairly narrow SDN or NFV project and they take off. Others have either not yet found the right target, or need something big and broad in scope. Picking the right target is important, in short.
Crossing the barrier
If you think focusing on the highest cost is the best approach, then you have to deal with the fact that cost management means working on interdependent processes that will increase the scope of your project or product. For example, total process opex costs across the six categories represents almost 27 cents of each revenue dollar. There are a lot of interdependent systems, practices and technologies tied up in those costs, and to address any single cost category might implicate half of all opex systems and practices, or more. So, you're back to boiling the ocean.
Perhaps a smaller body of water is in order. For some operators, focusing on opex change would make a lot of sense. Every process opex cost category is projected to show a cost increase through 2020, but some show a lot more than others. For example, one class of operator faces opex growth of 25% in only four years in one of the process opex categories (customer technical support). If you're a vendor trying to help a prospect justify NFV, a strong story in managing customer technical support costs would resonate for at least some of your prospects.
Remember, too, that the process opex categories tend to be interdependent. Customer technical support is only one of three opex categories that are growing as a percentage of total opex. Network operations expenses and IT service operations expenses are the other two, and in 2016 these represent a whopping 47% of process opex and 13 cents of every revenue dollar. Since fixing any of the three is clearly linked to software-driven automation of management and operations processes, why not target all three?
Answering the "How?" question
Both SDN and NFV need a new paradigm for software automation of their own service processes, and everyone in both communities and in the OSS/BSS area realize this. Can this new software automation paradigm be helpful in managing process opex? Could we extend the network abstractions of SDN and NFV to cover legacy elements, and by managing these new abstractions then extend software-automated operations management to legacy things too?
If the answer is "Yes," then we might open up a pathway to build early, small-scale, trials and services for SDN and NFV into something with meaningful impact. Vendors could target fairly specific applications and services, but use those early targets to pull through operations and management automation on a broad scale.
A specialized vendor like an MVNO or a managed service provider that has relatively little infrastructure to manage might easily build an SDN/NFV service that addresses a lot of cost with little more than CPE management. For mainstream multi-service operators it would be easier to do this if the early targets were big in themselves, such as IMS/EPC for mobile operators or perhaps a revolutionary vision of IoT.
Big targets mean big vendors, which often means proprietary solutions or silos. Operators usually try to address this through standards, but there's been little progress in establishing operations and management standards for either SDN or NFV, at least not standards meaningful to network operators. Open strategies and open source offer hope particularly if they can be used to build partnerships among vendors, but most such partnerships to date stop short of the operations layer.
Many operators think that because NFV could target multiple services and benefits, and because of the need to address costs and revenues quickly, silos may be inevitable. The easiest solution might be to focus on how to unite the silos, an approach that has recently been described as "choreography."
Choreography may be a new name, but it doesn't have to be a whole new software project or standard. It could be implemented using the same orchestration, automation tools and modeling used in the traditional SDN and NFV part of the implementation. And perhaps both the SDN and NFV communities should unite on making it possible.
— Tom Nolle, President/Founder/Principal Analyst, CIMI Corp., special to The New IP