These programs perform the previously manual tasks reliably and repeatedly without human input. This saves time for network engineers and reduces network outages caused by human error. It can even improve performance over time via machine learning.
Explain it like I'm five: Network automation means computers do things in networks instead of people. Over time, computers can learn how to do this better by themselves.
Why automate networks?
Network automation is beneficial for three main reasons:
Cost: Network automation lowers operating expenditure (opex). Fewer resources are required to achieve the same or better performance.
Performance: Network automation also improves performance and resilience. Many network changes are done manually, which increases the risk of mistakes and misconfigurations. Plus, many of the network tasks that can be automated could be considered the "boring" part of a network engineer’s job. By automating a network, engineers and architects can use their time for more impactful tasks. These can include network strategy, redesigns, or creative deployments.
Scale and efficiency: Network automation collects lots of data, and this can be used to improve performance in a way that was not possible before. Automated provisioning and changes to networks can also cut deployment times from months to minutes in some cases. This gives operators more time to make strategic improvements to their networks.
How does network automation work?
Network automation is a logic flow – ‘If X, then Y ’. So a network can be programmed to respond to certain triggers by performing certain actions. For example, a command might re-route network traffic along a certain path if performance along its usual path falls below a certain level.
Tasks that can be automated in this way include SLA (Service Level Agreement) monitoring, configuration changes across many networks, and diagnostics.
At a more advanced level, networks can be built with a centralised control plane that allows common network tasks to be templated, deployed, and adjusted when necessary, either automatically via machine learning or by network engineers. These are known as software-defined networks.
At the far end of the scale, full closed-loop automation is able to not only run a network without human input, but also improve it and learn from the diagnostics data generated.
In summary, there are three main types of network automation. These exist along a continuum:
Script-based network automation: This involves writing code to automate individual network functions or tasks.
Intelligent network automation: One step up from script-based automation, this is an administrative portal or software where companies can manage processes without needing to write commands manually. This involves deploying templates or programs to manage specific network functions.
Intent-based network automation. This starts from the required outcome – what does the user or customer want to achieve? - and works backwards from there. Intent-based network automation makes heavy use of artificial intelligence and machine learning.
What are some examples of network automation?
Etisalat (now e&) deployed an end-to-end multi-vendor DWDM transport network orchestration using Blue Planet software. This allowed automation of order-to-service processes, which cut deployment time by between 70% and 90%. Read more
Swisscom used Red Hat automation software to automate management of 15,000 network components such as services, firewalls, and storage devices, saving 3,000 hours per year of previously manual tasks. Read more
