Info Image

An AI-Centric Future: Driving Cloud-Native Networks of Today and Tomorrow | Juniper Networks

An AI-Centric Future: Driving Cloud-Native Networks of Today and Tomorrow | Juniper Networks Image Credit: The Fast Mode

The rise of big data, machine learning (ML) and automation is seeing artificial intelligence (AI) emerge as a critical capability across every industry. AI enables enterprises to leverage insights from internal and external data sources to predict events, automate decision making and enhance enterprise responsiveness towards fast changing customer needs. This creates a more agile environment where businesses become highly adaptive to market demand, with this greatly enhancing their monetization opportunities and long term sustainability.

Data-driven industries are essentially at the forefront of AI adoption and this includes the IT and networking industry where gigabytes of data is generated and stored every day across various points in the network. The use of AI in this space was discussed at the recently concluded Network Virtualization and SDN Asia 2019. At the event, The Fast Mode spoke to Nitin Vig, Chief Architect, APAC Center of Excellence (CoE) of Juniper Networks on how AI is powering the networking industry, in particular on how AI will be driving cloud-native networks of the future. The event which took place from the 11th-12th of September at the Sands Expo and Convention Center, Marina Bay Sands Singapore is part of the larger TechXLR8 Conference which showcased various technologies in the networking and 5G space.   

Nitin is an architect at Juniper Networks with 19 years of industry experience. He works closely with service providers and enterprises in the APAC region in areas of fixed and mobile broadband, security, SDN/NFV architecture and next-generation core networks. His areas of interest include DevOps, automation and machine learning. He holds a Graduate degree in Electronics from University of Mumbai and an MBA in Information Systems.

According to Nitin, proliferation of IoT networks and the growth in cloud-native deployments is seeing networking extending beyond enterprise and service provider core networks to the array of devices and services inside the Cloud. With networks becoming larger and more complex, network visibility becomes increasingly critical. This calls for more robust network analytics and this is where AI comes into the picture – as a critical capability needed to decipher all these information, predict future events and provide timely recommendations necessary for automated policy response. “AI helps to make sense of all the data that is being generated from inside the network so that we can absorb that data and deliver analytics to make networks more efficient, smarter and more resilient, and use that analytics to drive network behaviour towards a self-driving network,” says Nitin.

The shift towards a software-centric approach

The use of AI is driven partly by the continuous advancements in networking technologies. This includes the shift to a software centric approach and the widespread adoption of open standards. On the former, Nitin said that, “Software centric approach is essentially how you move away from traditional box-based rigid models and move towards models which are more dynamic and that are able to bring in services and turn them on in a matter of minutes rather than weeks or months.”

Open standards

On open standards, Nitin said, “The adoption of open standards in the service provider environments or even the cloud, and the idea of disaggregation and white-box were initially driven by cost savings and capex constraints.” However, over time, the value of open standards grew beyond monetary objectives to driving a bigger value, which is allowing network operators to take control of their networks. This includes freedom from single vendor lock-in, the ability to launch new services at a much faster pace and the freedom to adopt newer technologies.

Open source has always been something that Juniper Networks has focused on since inception, but in the last few years, the company has been doubling down on this approach. Tungsten Fabric, Juniper’s open source project is one such example upon which the company launched its Contrail SDN Controller. “Tungsten Fabric is now part of the Linux Foundation and is one of the fundamental technology for SDN,” highlighted Nitin. Apart from that, Juniper is collaborating with the Open Networking Foundation (ONF) and pursuing hardware certification for open compute projects. These are part of the many other different initiatives Juniper Networks is working on in this space. “In some cases we are leading and in some cases we are making sure we are building the ecosystem that is required,” added Nitin.

Elaborating on how open source drives cloud native networks, Nitin emphasized that the main goal of cloud native networking (via processes such as disaggregation, microservices, containerization and agile DevOps-based strategy) is centered around network agility. Open source essentially increases the control of the network, allowing services to be deployed as and when needed, and minimizing dependency on vendor ecosystems. 

How AI delivers network agility

It is this goal of agility that is pushing the use of AI as a critical capability within today’s networks.  Information collected from various data points across the network – mobile handset, radio, core or the data center –  is expected to go beyond graphical analysis to being able to help with network management decisions such as timely identification of issues (example, network bottlenecks) and timely deployment of new services. This is where ML and AI come into play, enriching network visibility with predictive analytics. By combining both internal and external matrics, AI is able to, for example, single out external factors such as traffic surge from a football match, from network-related factors to identify the underlying cause of network congestion. With further correlation of data, AI is then able to recommend the best policy response to address the issue in a timely manner while maintaining the overall user experience.  

Nitin also shared on how Juniper Networks is intensifying its effort in AI by incorporating the capability across its portfolio of solutions. Network analytics embedded with ML and AI are being deployed for example, in the cloud space, in its Contrail SDN Controller. This allows optimization of the cloud for certain use cases. AI is also being deployed across its security solutions. “On the security front – for both enterprise and service provider, (the question is) how do I leverage cloud to be able to prevent zero day attacks, respond faster to attacks I have not seen before (and) how do I take care of this terabyte-based distributed denial of service(DDOS) attacks that might be happening inside the network?” elaborated Nitin. The use of AI and ML enables Juniper Networks’ security solutions to detect malicious or suspicious traffic patterns and institute the matching security policies. On solutions for service providers, Nitin says that the telemetry collected from the network when combined with AI assists processes such as root cause analysis conducted through correlation of various data collected over time.

For enterprises, Juniper Networks’ recent acquisition of MIST is seen as a major move to enhance its AI offerings. The acquisition will enable Juniper Networks, according to Nitin, to work towards their goal of reducing the complexities of managing networks and increasing network efficiencies. The acquisition will enable Juniper Networks to extend this goal beyond SDWAN to cover campus, the local area networks (LAN) and the entire SD Enterprise networks. “MIST fits nicely into the whole vision, while it brings wireline network portfolio for us, the key is the AI technology it brings,” emphasized Nitin.

Asked to elaborate on what customers expect when it comes to the deployment of AI, Nitin said that there is mixed reaction from the market. On one hand, there is some apprehension that comes with giving away control of the network to ML algorithms. In this aspect, a gradual approach is being pursued with AI being deployed as a tool to push continuous improvements in network performance and for making networks more efficient. “Customers are still looking at ways to change the traditional way of operating the network, towards making them more self-driven using technology,” says Nitin.

Customers however are also greatly interested to leverage AI to enhance the value they deliver to their customers. Towards this end, Nitin said that AI has to provide real-time actionable data, for example in retail, when a customer walks into the store. In this scenario, AI allows enterprises to tailor more contextual offerings based on past data from the customer and external information sources, allowing businesses to make a difference in what they deliver to their customers. - end -

For more video interviews with Juniper Networks at the Network Virtualization and SDN Asia 2019 on topics ranging from SDN, cloud native networks, artificial intelligence to network security, visit the following:

 - How SDN Enables Enterprises to Meet Demands of Cloud-Native Workloads in a Multi-Cloud Environment | Juniper Networks

 - Artificial Intelligence to Enhance Agility and Responsiveness of Enterprise Networks | Juniper Networks

 - Telco Cloud Deployments, Challenges and Opportunities in the 5G Era | Juniper Networks

 - Juniper Networks at the TechXLR8 Asia 2019

 - Building the Cloud-Native Network | Juniper Networks

PREVIOUS POST

Taking 5G to the Max with mmWave

NEXT POST

The Evolution Towards the Right Network Cloud