Intelligent orchestration through a single orchestration framework or a series of orchestrators will be critical to bringing operational simplicity to the complicated mess of networks, systems, functions, services, orchestration and management technologies. After all, failing to consider complexities from the outset can mean increased operational costs, and diminished agility and time to market in an on-demand world.
Delivering the experience
Today, enterprises and consumers crave control and visibility into their own services. They want a device-driven and extremely personalized experience in which they have dynamic control over spend, account, service, groups and device types. They want to move toward an "order-for-one" experience in which they control their own offers, terms and parameters.
At the same time, CSPs recognize that the customer experience must become more personalized as well as all-encompassing -- the time-and-space barriers of personal life and work life must blur in a "now" economy defined by the likes of Facebook, Instagram, Snapchat and WhatsApp.
However, the type of extensibility, flexibility and service velocity required to succeed will require extreme programmability -- something that can only be achieved in the cloud and with virtualization capable of simplifying network complexity and the ways in which resources and operations are managed.
That brings us to the New IP in which social, mobile, cloud and big data must converge. In the New IP, the data center must shine. It can no longer be an esoteric entity important only to back-end apps. Rather, it must become the front door to customers and their myriad devices and applications.
The momentum toward network functions virtualization (NFV) will enhance the need for robust data center resources and virtual infrastructure managers (VIMs), increasingly critical to breaking free of the limitations of physical hardware. The work in OpenStack, OpenFlow and the OpenDaylight Project is evidence of how important the data center will be to cloudification of telecom infrastructure.
Without careful consideration of how to orchestrate services from initial service design all the way through to data center operation, CSPs run the risk of adding too many layers of complexity, operational streams and costs, not to mention adding points of failure and security holes.
To achieve simplicity, implementation plans should be built around intelligent orchestration, which encompasses service design and delivery, virtual network function (VNF) and physical network function (PNF) management, and of course, coordination with data center resources.
Whether through one orchestration framework or a series of orchestrators, the goal should be operational simplicity -- no matter the complexity inherent in connecting VNFs and managing virtual machines, software and processes.
In order to maintain visibility and control over networking capabilities; resources utilized by different services and the critical linkage to OSS/BSS (e.g., ordering, fulfillment, billing, etc.), orchestration must cover:
- An application orchestrator provides a rapid path to NFV by managing all capacity and virtualization coordination for virtual network functions. CSPs can free themselves of the manual configurations that make managing network functions and their lifecycles slow and error prone. Rapid responses to changes in network capacity requirements are achieved by monitoring key performance indicators (KPIs). It can also integrate with multiple virtualization infrastructure systems to include other vendor environments.
- A network services orchestrator coordinates network services and would integrate with the application orchestrator. With it, CSPs compose and deploy network services quickly and within a standardized framework, managing the lifecycle of network services across different vendors’ products and across diverse locations from a single application. It also ensures network services meet customer demands by elastically scaling them and automatically updating service connectivity.
- NFV infrastructure (NFVI) which includes the data center resources that are managed using virtual infrastructure managers (VIMs) -- cloud management systems tasked with managing the virtual machines that run on hardware in the data center. The NFVI is the foundation for infrastructure management and communication with NFV analytics which offers insight into customer actions and network resource utilization.
In seeking out solutions with the above components, CSPs should consider three requirements:
- Network elements released to VNF (e.g., session border controllers, signaling routers, EPC, etc.)
- High-volume analytics and policy: These are absolutely critical pieces in managing the interplay among OSS/BSS, EMSs, VNFs, NFV instances, NFV orchestrators and VIMs
- Cloud computing and data center technology expertise: Policy rules that govern network and service behaviors must converge with analytic feedback from run-time operations so that CSPs can orchestrate the many streams of communication necessary to bridge physical and virtual environments
The common thread here is the value of analytics. By infusing a high degree of intelligence through business and technology rules and with an innate knowledge of resource availability, CSPs can then use that intelligence to dictate how workloads are managed and how decisions are made around certain key metrics and measurements.
So why does this matter? The intelligence derived from an analytically charged network will help CSPs cultivate insight into network and customer behaviors that ultimately drive QoS decisions, new offers and network resource allocations. As that happens, CSPs can more readily understand the many ways in which virtualization affects core network functions, such as provisioning, service quality management, network performance and the overall customer experience.
It then becomes easier for CSPs to better manage how network assets are utilized and how data traverses OSS/BSS and service layer platforms -- even as more service instances, physical network function instances and VNFs come into play.
And last but not least, it becomes easier for CSPs to better monitor and acquire vast stores of operational data (which must be evaluated against billions of transactions and events), as well as feed reporting information about network elements, servers, databases, applications, OSS/BSS and customer support.
Only with this type of intelligent orchestration can complexity be conquered and simplicity realized in the New IP.
— Barry Hill, global head, NFV, Oracle Communications, special to The New IP