The drive toward virtualization with NFV and SDN for networks everywhere is leading us down a path that will eventually lead to the creation of our own form of artificial intelligence. True to that trajectory, the management and orchestration systems necessary to optimally operate the next generation of virtualized networks will have all the characteristics necessary to be considered self-healing, self-aware, and self-evolving.
Entering the Age of Virtualization
In October of 2012, mobile service providers jumped head first into the world of virtualization with the release of the seminal ETSI (News - Alert) NFV whitepaper. Mobile service providers look to leverage the virtualization movement by leveraging a common hardware and management architecture for disparate core data plane packet processing and control plane functions.
The benefits of NFV include increased speed of time to market and services that can be rapidly scaled up/down as required. In the virtualized and cloud computing world, this means that mobile service providers want to achieve agility and elasticity. Agility is the ability to add, remove, or change services quickly and efficiently. Elasticity provides on-demand resourcing, making services scalable to meet the network demands based on capacity, location, and function.
It is interesting that mobile service providers created NFV based on virtualization and cloud technologies that still exist in enterprise networks today. But, because the NFV model is focused on virtualizing network services vs. endpoint applications, the enterprise businesses have found value to bring the NFV architecture back into their own virtualized SDN and cloud environments.
Businesses are continually developing more complex and robust management systems to provide visibility into the network elements and allow operational staff to make adjustments based on the information received. As these organizations move toward a virtualized core network infrastructure, it is becoming essential that the management system deliver a complete holistic view of the entire network and have the capability to automatically make the operational adjustments necessary to keep the network running at maximum efficiency.
Currently, we are focused on use-case and function-specific NFV and virtualization management and orchestration scenarios. The networks are comprised of different components and technologies such as the firewalls, application delivery controllers, application gateways, and other hosted services. There are isolated situations where these virtualized architectures are being tested to deliver a specific function or capability. DDoS mitigation or highly available videoconferencing services may be the short-term use case where businesses can extract immediate value.
It is obvious that what occurs at one component such as the firewall affects multiple application services and other components. At a technical level, consider voice over IP. A VoIP call interacts with the SBC, SIP directory server, SIP application server, application firewalls, network firewalls, and more while using protocols and technologies including SIP, RTP, DNS, AAA/RADIUS/Diameter, and others. There are probably very few components within the network architecture that are not essential for the delivery of a VoIP call.
All of these individual scenarios being utilized today will eventually need to integrate into a master management and orchestration infrastructure. Ultimately, there is an interdependence that needs to be acknowledged and addressed as we move forward.
It is essential that the MANO understands the interactions of the different components even though they may seem isolated and the type of analytics collected may be different. MANO solutions need to evolve to incorporate the different services and technologies within the entire network architecture. This means that an intelligent heuristic engine must understand the interactions and interdependencies across these functions and protocols. Like individual people, every network is somewhat different, so each MANO solution must have its policies and processes adjusted accordingly.
Management and orchestration systems are critical to the success of these virtualized networks. The continuous changes in network functionality and network service forwarding graphs based on the agility and elasticity of the environment require speed and efficiency that is not possible at the manual level when human intervention is required. Only a fully integrated and automated management and orchestration system – with the ability to analyze data with heuristics based on defined policies, and then interact with the network ecosystem to generate changes that maintain the network functions within defined thresholds – can support the next generation virtualized networks efficiently and effectively.
Edited by Stefania Viscusi