Intent-based networking offers significant advantages to modern infrastructure teams. But how close are we to the self-operating network? Mansour Karam, CEO of Apstra, unravels the rise of intent-based networks and explores whether we can yet trust the most advanced networks to apply their own corrective actions.
‘Intent-based’ increased in popularity in 2018, but how much is real? Intent-based networking is a major advancement towards the self-operating network (known as level 3 intent-based networking), but it is itself made up from many smaller advances.
It has many pretenders offering partial features such as basic automation (level 0 intent-based networking) while missing support for vendor-agnostic infrastructure, single source of truth and the ability to deliver real-time change validation regardless of underlying hardware or operating system technology.
The first significant feature of intent-based networking must be an architecture built on a single source of truth, embracing both the intent and the actual network status – including control of every aspect of the network service
Without this single source of truth, questions like ‘Which users will be impacted if link x fails or gets congested?’, or ‘What is the link utilisation on all links carrying user X’s traffic?’, mean consulting the network map, checking that it is up-to-date, checking the operational state of every link – and so on and so on.
Even when all the necessary data has been collected and collated, the final answer still has to be calculated. Whereas a network with a single source of truth has all the necessary data in that one source and, in a truly intent-based system, the answer to those questions can be automatically and speedily calculated and displayed.
This single source of truth has already been realised in the most advanced data centre automation solutions and it marks a very significant step towards the self-operating network or data centre.
Root cause identification
Another vital step is ‘root cause identification’. Once you have a single source of truth encapsulating both what the network should be doing and what it is actually doing in real time, then further intelligence can study the incoming telemetry, note any irregularities, outages or performance degradation and not only report them but also trace them back to the underlying root cause.
Whether it is cars, computers or domestic appliances, there is nothing more maddening than an intermittent fault. Something is wrong but it all works perfectly when taken to be serviced.
Whereas the service operator studies the performance during inspection, a proper intent-based system keeps a full ongoing record of when and how things went wrong, together with the full knowledge of the infrastructure and component parts.
The combination of a single source of truth plus AI can trace back from erratic symptoms to pinpoint the root cause.
The majority of data centres are adopting leaf-spine or ‘Clos’ network topology to provide better application performance. The traditional three-tier hierarchy generated a lot of traffic up and down the system, while the ‘flatter’ leaf-spine topology places hosts at equal distances, and applications behave more predictably. The result is a higher bandwidth, lower latency, while relatively easier to manage network.
‘Relatively’ is the key word here because, even for small leaf spine networks, network administrators have to manually track and verify thousands of logical and virtual elements, as well as physical components.
If just one of these is wrongly configured or fails, a cascading effect can affect multiple compute, network and storage nodes with unpredictable consequences for system performance, user experience, business – and ultimately reputation.
If, for example, the network operator needs to remove or adjust a working switch in the middle of the business day, an advanced intent-based network will allow the traffic to bypass that device in such a way that everyday business can continue while the switch is changed.
When the device is restored, the system monitors will be immediately aware, so that system visibility and performance monitoring continues uninterrupted. All this can be managed in real time from a single management console and simple graphical interface.
It is easy to see how this centralisation and automation of management saves a lot of time and labour, as all those relatively fiddly and boring parameter settings are automatically and precisely rolled out across the network. But in practice it is the reduction in human error that provides the biggest saving.
Trusting networks to run themselves
The network operator’s role is becoming less like a motor mechanic and more like a Formula 1 driver, with a clear view of the road, the vehicle’s position, speed and remaining fuel – all that is needed to fulfil intent – without being overwhelmed by technical data about the system’s inner workings. The IT team can now meet deadlines, guarantee performance levels, specify upgrade requirements and speak the language that business understands.
We can now rely on the most advanced intent-based networks to identify the root causes of problems, but can we yet trust them to apply their own corrective actions?
How soon that will happen in the real world is as much a question of the operators’ level of trust, as it is about the technology itself. What would it feel like to take a first trip in a driverless car or aeroplane?