The major topic of discussion at Nokia's Network Management Breakfast, held in Nice during TM Forum Live 2016, was automation. Important questions were addressed: What's driving the automation of network operations today? What are the challenges and best practices? How can operators achieve the extreme automation needed to support 5G goals in future?
Participants acknowledged that competition around the speed of service delivery and a strong emphasis on customer experience are driving their interest and investment in automation. They agreed that the adoption of a customer-centric mindset focuses an operator on automation because it needs to understand customer experience of services, which is at a completely different level from managing lights on boxes. Operators typically employ hundreds of people in their network operations centers (NOCs) watching hundreds of screens, but can't map their actions at network level to improvements or otherwise in customers' experience of services. Joining the dots between the network and services requires the transition from NOC to service operations center (SOC) and the automation of network/service/customer data collection, its correlation and analysis.
As part of this journey, what's meaningful to customers about the service experience needs to be analyzed and the right key performance indicators (KPIs) extracted and modeled. Network engineers need to be upskilled for their new SOC roles. Such a transformation requires operations staff to be involved from the beginning if it is to be successful, and human resources also needs to participate to help define the new roles and training programs. The cultural and organizational implications of bringing in automation are extremely important as operators need to allay fears and take engineers with them. Such automation flattens the organization and opens it up to new, real-time customer interaction models: The SOC owner's operations staff communicate with customers directly as soon as they spot a service impact rather than inform the contact center, cutting out cost and delay in reaching subscribers and improving customer experience.
Automation is expensive and there is always more to do as new technologies are brought into the network and new automation tools become available. Automation is not a one-off task and technological obsolescence is a real concern. Participants said they would like to have common automation tools that they could use across different domains of the network to start breaking down the silos that exist right now. This kicked off an interesting discussion around the idea of a common automation framework through which engineers and programmers could make automation artefacts available for reuse. Internally, participants thought they might prefer to standardize on a couple of common languages and data model approaches for automation, but they were also intrigued by the idea of an open industry-wide automation platform from which they would be able to retrieve automation fragments, perhaps a script for a specific piece of network equipment, for example, to save the effort of having to write this for themselves. Participants agreed that there is a difference between automation artefacts that have no competitive value and those that touch the business, for example, that are a source of competitive differentiation. The former would be good candidates for inclusion in an open automation platform.
Concerns were raised over the possibility of a "two-speed" organization created through automation. The virtualized network is likely to be more highly automated than the traditional, physical network, for example, which may cause operational challenges. One participant pointed out that 80% of his NOC activity is caused by pre-LTE technologies because of the higher levels of automation with LTE: This is cost he would like to cap, but at the same time, does it warrant investment in automation when these earlier network technologies are on the way out?
Automating problem resolution -- root-cause analysis -- was seen as another major challenge, although one that participants felt could be resolved with more data. There was strong consensus around the basics of the next generation service assurance architecture needed for a virtualizing network and by operators that want to become more customer-centric and data-driven. Such an architecture is based on a common big data platform that collects data from numerous sources, including data on customer service impacts, customers, the network, planning, even social media. Analytics play a big role in the architecture and participants are particularly interested in moving to predictive analytics as a means of driving automated feedback loops around capacity management, QoS maintenance and root cause analysis. As one participant put it, he wants an assurance architecture that provides a seamless view from the customer down to components and he wants it to be proactive so that he's never in a position where things go wrong.
Participants unanimously assumed that virtualization will be the foundation of their networks in the future. They also thought they would have limited control over their networks -- their customers will be in charge and will dictate what they want from the network. This resonates with the idea of network slicing, which is a key concept in 5G networks, alongside other concepts including ultra-dense networks, super-low latency, the connection of things as well as people and 50 billion devices accessing the network by 2020. All these 5G attributes will push automation even higher up operators' agendas than it is today as the scale and real-time nature of 5G operations simply won't be possible without automation. Operational data will be far more distributed across the 5G network, confirming the trend towards big data and analytics-driven management -- and the network will need far more intelligence at the edge, especially if it's supporting new IoT use cases.
However, participants pointed out that to get to this holistic vision of operations, strategy planning and procurement in their organizations need to change. Network components -- physical appliances today -- are still selected for their functionality, not their operational implications. There needs to be far more recognition of and focus on the importance of managing the network and cross-organizational funding of its automation.
Altogether, this was a fascinating and wide-ranging discussion of multiple aspects of network management and the critical importance of automating it, starting today but with tomorrow's 5G networks in mind. Each participant brought best-practice examples to the table and was an inspiring source of insights and views.
This blog is sponsored by Nokia.
— Caroline Chappell, Practice Leader, Cloud & NFV, Heavy Reading