Network functions virtualization isn’t only about moving functionality out of purpose-built hardware into more flexible software architectures; it also means a monumental change in the way service providers manage the lifecycles of products and services. Nancee Ruzicka takes a look at what needs to happen.
In this new era of digital services all traffic is data and all products are in fact applications. The access and transport network infrastructure is no longer unique to each service offering. Applications with specific features and functionality are combined with basic connectivity to create a unique product that is delivered over any type of wired or wireless access network. This fundamental shift in product lifecycle management (PLM) is profoundly changing the way service providers operate their businesses.
Rapid introduction of new services isn’t a revolutionary idea; it’s been a priority for service providers since networks began converting to IP and mobility was brought to the masses more than 20 years ago. And in survey after survey, service provider executives indicate they are considering network functions virtualization (NFV) as a way to improve time-to-market for new digital services. (Learn more about the results of TM Forum’s survey on NFV trends in our Insights Research report When will NFV cross the chasm?)
NFV isn’t enough
NFV alone can’t solve the problem. In fact, service providers cannot effectively utilize NFV without sophisticated and automated orchestration. That same orchestration and automation must be applied to new PLM processes so that the product components subsequently bundled into digital services can be rapidly defined and introduced to the product catalog.
Given the volume of virtual instances likely to occur in the coming years as NFV is introduced, implementing a dynamic PLM strategy is essential to developing, delivering, managing and monetizing digital services. Over the past decade, service providers have been adjusting to this new reality and transforming operations to implement centralized product catalogs. These include, or are tightly integrated with, the service and resource catalogs necessary to define products, configure products into digital services, monitor performance and retire products that are no longer viable or valuable (see Figure 1).
Introducing NFV to PLM
When introducing NFV to the product lifecycle, the ideation and definition stages remain relatively unchanged, but design of new product components becomes more involved. In a catalog-driven architecture, granular rules and workflow are included in each component to execute offer management, order management and provisioning. Components must accommodate configuration of virtual elements – and current non-virtualized components in a hybrid environment – and respond to policy-driven adjustments as the network is scaled and optimized.
By making required adjustments in the catalog and service-delivery layer, rather than in each product and service, service providers are able to dynamically adjust their products. This feature is critical if they intend to rapidly bundle and deliver new complex products that require configuration and provisioning of NFV, physical network infrastructure and cloud-based applications. TM Forum’s Service bundling in a B2B2X marketplace Catalyst demonstrated this capability. For more about the Catalyst and the new Yellow Pages-like catalogs that TM Forum’s Zero-touch Orchestration, Operations and Management (ZOOM) team is proposing as part of its NFV Procurement Survival Kit.
Component-based products that are driven by thresholds, rules, policies, automation and integration are foundationally more agile, flexible and extensible. In addition to reducing the time required to configure and bundle services, reusable components and automated processes reduce the cost of launching and retiring products.
As the number of products increases and shelf-life shrinks, service providers can no longer justify long design, configuration and roll-out cycles. Likewise, the sheer volume of products possible using virtual components and cloud applications means that the catalog can be continuously groomed and optimized. Another TM Forum Catalyst called Preparing NFV for Primetime demonstrated this optimization through a metamodel for event-driven management and operations. It included user interfaces to provide role-based views and policy-based management for security, optimization, compliance and governance.
Monitoring products that incorporate both virtual and physical components requires additional vigilance and intelligence. In the TM Forum Catalyst Maximizing Profitability with NFV Orchestration, which was first discussed at Digital Disruption in San Jose and is continuing at TM Forum Live! 2015, the participants are using analytics to monitor product performance and dynamic policies to optimize a hybrid network.
In addition to applying analytics and intelligence to real-time performance monitoring, optimization is accomplished by adjusting the product components in the catalog. In some cases these adjustments reflect economic metrics, such as computing or transmission costs, as well as network performance. Catalog-driven service orchestration and fulfillment then rapidly respond to workflow triggers and adjust the service infrastructure.
The service instantiation, resource-matching and scaling demonstrated in the Catalyst becomes a significant part of the product lifecycle when NFV is introduced. Real-time monitoring and analysis of product performance is what drives policy-based adjustments in the physical and virtual network. That same analysis is important feedback to the PLM process and helps service providers understand when products should be adjusted or retired.
The volume of product components and service combinations increases by an order of magnitude when NFV elements are added. Mixing and matching components to build a new product requires fully automated processes that include pricing rules, economic metrics, configuration policies, workflow and monitoring thresholds.
Automation applied to the PLM process delivers the orchestration required to rapidly generate an offer, configure product components into a reliable digital service and deliver that service to a customer. But that orchestration cannot be accomplished without complete end-to-end automation of the underlying operational and business support systems (OSS/BSS). Automation of the PLM process is essential to define functions, products and services in ways that enable rapid instantiation of digital services – both physical and virtual.
Absorbing layers of virtualization into the PLM process takes time, and in addition to the difficulties experienced when trying to monitor virtual network elements, other challenges are cropping up, including:
- OSS/BSS interoperability – service providers have made large investments in operational transformation and OSS/BSS upgrades and replacement. Those solutions are used to execute the PLM process and will remain in place. Each must be expertly configured and augmented to incorporate fulfillment, assurance and billing functionality to engage smoothly with PLM orchestration and support NFV deployments in a hybrid network and between partners’ networks. TM Forum’s ZOOM team is working on these issues as part of its Frameworx 15 deliverables.
- Poor processes – service providers worldwide recognize they need to automate important and complex business processes. Adding and managing hundreds of new NFV elements is not possible without automated workflows, but automating a bad PLM process will not add value. Rather, it will result in an increasing number of errors that need manual intervention which adds cost and damages customers’ experience.
- Data, data, data – the volume of data being collected and processed by service providers continues to grow in volume and complexity. Adding NFV components increases the number of elements included in product, service and resource inventories exponentially. Every product component, configuration, customer activation and transaction must be captured, correlated and acted upon. Collecting, disseminating and initiating PLM actions, given all the data available to service providers, is overwhelming. They will need increasingly granular monitoring and rapid response to support NFV.
One of the core tenets of virtualization is automation. To make widespread virtualization work operationally, PLM processes that are currently a combination of OSS/BSS and manual tasks must be entirely automated. Successful integration of NFV into PLM processes will rely on catalog-driven architectures to enable seamless execution of business processes. This is why many operators are transitioning to strategies that go beyond business process and workflow management to centralized, data-driven PLM orchestration.