As developments like artificial intelligence, data-driven analytics, and machine learning work their way into enterprise infrastructure, some experts are starting to wonder: how long will it be before we see the autonomous, self-correcting network?
Not long, actually, if some of the research and development projects already underway start to bear fruit.
According to IT Buyer's Resource, leading platform developers like Cisco and Juniper are already building autonomous network management capabilities utilizing many of the same technologies going into autonomous vehicles. The goal is to devise a management stack that can configure, monitor, and maintain the network environment with little or no human intervention, leading to a world in which users or even applications themselves can merely state their requirements and have the optimal network configuration delivered to them at a moment's notice. Not only would this provide highly customized data environments, but it would also vastly improve network security and drive down the cost of data services.
On an operational level, this is called intent-based networking, says Gartner. With intelligent management software, networking ceases to be a product or even a market but an event designed to achieve an objective. To get there, the management stack will need to embrace four key elements:
- lTranslation and validation to convert network policies into network configurations;
- lAutomated implementation and orchestration to extend configurations across multiple topologies;
- lState awareness using real-time status monitoring that is protocol- and transport-agnostic; and
- lDynamic optimization and remediation to maintain optimal performance levels.
Autonomy is also becoming a hot item on wide area networks and the 5G infrastructure that is evolving around Internet of Things (IoT) applications. Design firm Aricent recently announced new enhancements to its autonomous network solution (ANS) that aim to turn today's networks into intelligent platforms capable of supporting smart cars, smart cities, and all the other smart things currently in the works. The system now incorporates management and orchestration over virtual network service chains, policies, and performance metrics, as well as intent-based service lifecycle management and a self-optimizing framework that features predictive defect detection, root cause analysis, and support for low-latency services. The company says this will allow networks to convert raw input into direct commends to support more immersive user expe riences.
Network autonomy will also be a key factor in the development of autonomous agents that are poised to take over many of the rote tasks of our daily lives. As virtual personal assistants like Siri and Alexa become more prevalent, networks will have to adapt to agents' way of operating rather than the relatively cumbersome aspects of human control. This will require networks to support seamlessly the myriad interdependent functions that automated bots require, such as gathering, interpreting, sharing, and transferring of data. At the same time, organizations will have to overcome the silo-based infrastructure and single-vendor architectures that still inhabit many data ecosystems.
Autonomous applications navigating autonomous networks in support of autonomous devices – it seems that the future data environment will have a whole lot going on beyond our control. But this is not necessarily a bad thing given the enormous amount of time and energy that goes into manually performing tasks that technology can do better on its own.
With autonomy as a core networking capability, instead of people working to make a better network we will have a network working to make better people. – Arthur Cole