The hype machine for intent-based networking has been working overtime for the past 18 months.
Yet, despite all the hoopla and attention, many people are left wondering what intent-based networking (IBN) can actually do for them in 2019. In this article, we’re going to strip away the marketing noise to see if IBN should be on your radar — and, if so, where to start.
The main concept behind IBN is to look at the network from an entirely different perspective. Traditional networks were built using ample resources based on quantifiable metrics like throughput, latency and packet loss. Once the network was built to spec, applications were added on with little regard for the criticality of data flows. This resulted in a network that largely treated all data the same, with quality of service being the primary differentiator in terms of offering differing levels of network services.
IBN, on the other hand, is an attempt to look at the intended outcome of a data flow that a particular end user or business unit finds most acceptable. But even this explanation is a bit misleading. Removing even more marketing rhetoric, IBN is about the network having a clear understanding of the purpose, use and importance of all the applications it runs.
With that understanding, IBN uses AI to take this high-level overview of application service intent and convert that into detailed, end-to-end network policy. The IBN system then constantly reviews, updates and pushes this policy using real-time automation. Thus, all the underlying complexities are masked and managed using AI, as opposed to human network engineers.
Yes, IBN is likely the future of enterprise networking, but don’t expect your network to gain this level of intelligence overnight. IBN will require a great deal of planning and rearchitecture in your current environment.
Intent-based networking approaches
While IBN has been around for a while, most people really started to take notice when Cisco announced its Network Intuitive strategy at the company’s 2017 Partner Summit. Some vendors currently have platforms with end-to-end IBN, but only large organizations — think global service providers — are finding any sort of ROI in the early stages of this new technology. That’s why you may notice that most technology vendors pushing IBN expect their customers to ease into the new technology to gain experience.
Data analytics. Some IBN vendors are pushing the use of intent-based data analytics as the gateway to full-blown IBN. With this approach, network architects can begin to understand how application-specific data can be extracted and analyzed on their network to build and deploy business-specific network and data security policies. The idea is that once organizations understand how AI can analyze data flows to create a new policy, the next step would be to simply automate policy creation and deployment without human intervention.
Segmentation. A different IBN approach is to segment the network into intent-based and non-intent-based segments. This segmentation gives you the full IBN experience in certain critical areas where ROI can be easily gained.
At the same time, this approach reduces the upfront Capex and lessens the complexities associated with deploying AI across the entire corporate network. The two most common network segments in which IBN is being deployed are the WAN and the data center.
IBN in the WAN. The WAN is one area of the modern enterprise network where it’s impossible to simply throw more bandwidth at a problem. This could be due to limited WAN connectivity options or the high costs of providing increased bandwidth where needed. Instead of using more bandwidth to fix WAN connectivity issues, organizations can use IBN to help prioritize data flows in an automated fashion. Many companies are accomplishing similar goals with software-defined WAN technologies. IBN uses similar underlying principles, but it adds yet another layer of intelligence into the mix.
IBN in the data center. When IBN is deployed within the data center, the AI is focused more on server-to-server communication than on server-to-user communication. This focus can dramatically simplify configuring and maintaining data flow policy in distributed data center environments. Additionally, it has been shown that the potential for human error is reduced when AI creates and deploys new data flow policy.
It’s safe to say near-term, realistic implementation options for intent-based networking are relatively limited. Yet, despite these limitations, it is critical that businesses at least begin looking at different IBN-based options today to start planning larger and more meaningful deployments in the future. Organizations will need to take baby steps with IBN in order to get to a fully realized network that has the intelligence to route traffic based on their desires.