Nokia’s Event-Driven Automation (EDA): Driving human error zero

Imagine if your multi-vendor data center could be provisioned in a few mouse clicks, could automatically diagnose problems before they become outages, and could drive human-caused outages to zero. Nokia’s EDA platform is here and it’s tackling the biggest problems in networking.

Last week I had the privilege of serving as a delegate at Tech Field Day Networking Field Day exclusive with 诺基亚 #NFDxNokia, where they unveiled their EDA platform. Nokia built a vendor-neutral data center network automation platform that builds stable networks, and everything in the system is event-driven. The controller can determine state from streaming telemetry, and leverages state data to ensure the network operates as intended.

EDA is highly extensible for those who want to customize the solution, but plug-and-play for traditional network engineers. Its modularity is simple and intuitive, as demonstrated in their integrated “app store,” included in the platform. Its natural language feature makes it extremely easy to query the system, just like you would talk to a colleague.

What is networking’s biggest problem?

If you ask business leaders or industry analysts what networking’s biggest problem is, they’ll probably say “speed.” While other technology teams have leveraged automation tooling to increase speed for decades, networking continues to lag. Waiting weeks for a new VLAN is not unusual in traditional network shops. ?

The public cloud introduced a paradigm shift by transforming traditional static infrastructures into on-demand environments. The speed conversation goes something like “If we could just manage our on-prem with the agility of public cloud, we could take market share.” This thinking is largely what’s behind network automation’s velocity argument. So, is speed networking’s biggest problem?

Solving the correct problem.

Data center networks are complex, often fraught with technical debt, and routine network changes can devolve into break/fix marathons with little to no warning. As Senior Director of Product, Bruce Wallis reminded us, “2/3 of network outages are caused by humans.” Historically, organizations add layer upon layer of process to mitigate network outages caused by network changes, but this practice slows down the speed at which a new product or feature can be brought to market.

Another painful point was made by VP of Data Center Michael Bushong who said, “We have built a discipline where our stuff works best when everybody goes home and stops touching it.” He continued, “If we built airplanes like we build networks, I would walk everywhere and always be looking up for falling airplanes. What we have to address is the safety issue, not the speed issue.” Bushong’s point was well understood by those in attendance. Networks built by traditional methods and tooling are plagued with fragility and the only path to stability in that paradigm is month-long holiday change freeze windows.

Bushong added, “Does automation make things good? No, it just makes them more, so if you don’t have something safe to begin with, automation is the fastest way to break things at scale. We want to drive safety, so we can take away the fragility in networks and if networks aren’t fragile, they can be fast.”

How does EDA make networks safer?

Nokia’s EDA platform allows network operators to make changes to their network, without fear of breaking things. Their goal of driving human error in network operations to zero is accomplished, in part, by the following features.

Zero Touch Provisioning (ZTP): EDA automates the secure onboarding of new devices, reducing human error in device setup. With topology discovery and ZTP, new devices are provisioned quickly and securely, minimizing potential risks during the onboarding process.

Multivendor Support: EDA simplifies managing a multivendor environment by abstracting the complexities of different device configurations. This uniformity ensures consistent security policies across all devices, regardless of their manufacturer.

Intent-Based Automation: Instead of manually configuring devices, EDA uses high-level intent to define what you want to achieve, then automatically generates the necessary configurations. This reduces the chances of manual errors that can lead to security gaps.

Digital Twin: By using a digital twin to simulate network changes before applying them in production, EDA allows you to test and validate configurations in a safe, controlled environment. This ensures that potentially disruptive changes are thoroughly vetted before impacting live operations.

Pre and Post Checks: Before applying any new configuration, EDA performs pre-checks to detect potential conflicts. After the change, post-checks verify that the intended configuration has been applied successfully. This proactive approach prevents incorrect or insecure configurations from being deployed.

Network-Wide Transactions: EDA handles multiple configuration changes as a single, atomic transaction. If any part of the process fails, EDA can roll back all related changes, preventing partial configurations that could leave the network in an insecure or indeterminate state.

Revision Control: All configuration changes are versioned and stored in a Git repository, allowing you to roll back to previous configurations if necessary. This version control adds a safety net, ensuring you can always restore a secure state if new changes introduce issues.

Fabric Observability: Real-time monitoring of network traffic provides instant visibility into east-west and north-south traffic flows. This deep observability, combined with telemetry, allows you to detect anomalies and security threats as they happen, enabling a swift response.

Remediation: EDA continuously monitors the network's current state and compares it against the desired configuration. If discrepancies are found, EDA automatically corrects them, ensuring that the network is always provisioned as intended, minimizing the risk of misconfigurations that could lead to security breaches. The natural language model for interacting with the network makes it easy to get information from the network, regardless of the underlying manufacturer or device type. This should speed up issue identification and remediation.

By leveraging real-time event processing and automation capabilities, Nokia’s EDA platform transforms traditional static network infrastructures into dynamic, self-adaptive systems capable of handling modern-day demands efficiently. Unlike traditional static network management, which relies on predefined configurations and manual interventions, Nokia's EDA platform utilizes dynamic automation to make data center networks adaptive, responsive, and resilient.

Demo/Labs

The demo we were shown consisted of a clean UI, managing a small topology of 12 nodes. With 4 mouse clicks, a fabric was created with 586 “custom resources,” which mapped to thousands of lines of validated configuration. The simplicity of the UI made creating a fabric feel almost too easy, considering how much work occurred under the hood.

The labs were simple, made sense, and helped reinforce the key concepts we learned during the demo. We created a fabric with a YAML file, created a VLAN in the UI, mitigated a transaction failure the platform caught (VLAN already in use), and asked the fabric questions in natural language (show me interfaces with rizz.)

Take Aways

I walked away from this event with the feeling that Nokia listens to its customers, chooses the most impactful problems to solve, creates elegant solutions that are both easy to use and extremely extensible for the super nerds among us, and is a strong contender in the data center space. The EDA we saw is a v1 product, but it looked and felt like a fully capable solution. If Nokia’s huge success in the telecommunications carrier world is any indication, I’m excited to see what they do next in the data center space. I have high expectations for EDA’s next release.

I was equally impressed with the quality of Nokia’s solutions, not the least of which is their Service Router Linux (SR Linux) NOS, which boasts the most complete set of programmatic and telemetry interfaces in networking. Depending on how you measure NOS quality, SR Linux is orders of magnitude better than the NOSs of our industry’s most-known names. In my experience as a network engineer, NOS quality could make a big difference in the reliability of our networks.

EDA is an automation solution packed with cutting-edge features and a low barrier to entry. Stellar quality, a rich heritage of success in networking and solving the most impactful problems with some of the brightest minds is what Nokia is doing, and it’s compelling.

To learn more:

Watch the replays at the following links:

https://shorturl.at/tjZLd

https://shorturl.at/9ZHqD

Nokia EDA product page:

https://www.nokia.com/data-center-networks/data-center-fabric/event-driven-automation/