Insights & Resources

Network Automation - Smart Architecture (Part 1 of 3)

Network engineers are beginning to build complex platforms and create front and back-end systems to manage an entire network infrastructure.

Ian Reither
Smart architecture banner

Network Automation

"R2-D2, you know better than to trust a strange computer” - C-3PO
The public Internet was not purpose-built for the delivery of real-time communications. However, delivering real-time communications is essential for businesses, large and small, and it is being done through the adoption of modern software, network, and infrastructure paradigms. With the ever decreasing costs of these technologies, the industry has seen adoption from even the smallest businesses in the most remote parts of the world. An upstart operator can now leverage now generally available technologies to provide real-time communication services alongside industry incumbents using more traditional paradigms.
This article is the first of 3 posts that explore how ultra reliable and high fidelity real-time communications can be made possible through combining network automation, strategic peering relationships, and cloud agnostic points of presence. Whether you are familiar with these concepts or they are something that your company has not yet encountered, you will see why some of the most forward-thinking firms are driving adoption of and placing their trust in these architectures and technologies.
Network management has been going through an evolution over the past few years in which network operators and organizations are making automation a key driver in their strategy for building and managing their infrastructures.
While Software Defined Networking and Network Function Virtualization tend to receive the most attention, there are a number of obstacles that are making the transition difficult and slowing adoption. In the interim, however, there are intermediary steps that operators can take to begin the transition to more centrally managed and automated networks. One such step is the implementation of network automation such as Zero Touch Provisioning (ZTP).
A Zero Touch network implies the shift away from direct configuration via command-line interface (CLI) of network devices to the use of automation platforms. These platforms come in a variety of forms, but have the common goal of removing human interaction from the CLI by allowing the creation of deterministic environments through configuration management. With ZTP, organizations can pre-determine everything from baseline configurations to more complex configurations such as building configurations to implement a customer network that spans a global IP backbone.
This can ensure that something as simple as an interface description or a VLAN name follow the designed naming conventions. We can deterministically build everything without concern over whether a BGP peering was built correctly. Therefore, if your BGP peers should always “send communities”, then ZTP is your best bet in being 100% deterministic compared to an engineer who may mistakenly skip a line of configuration due to manual configuration from the CLI. After provisioning is complete, these same platforms can be used to maintain networks.
For example, using tools like Ansible, network engineers are beginning to build complex platforms and create front and back-end systems to manage an entire network infrastructure. Imagine that the source of truth of an environment now exists within your automation platform’s backend database instead of text files of configurations that may or may not be current and relevant.
Clearly, the benefits of this paradigm are enormous. Engineers can build platforms that routinely check for changes in the live environment and can take action against rogue configurations that do not meet implemented design. For example, automation in the system can detect if a network engineer modifies an Access Control List (ACL) that leaves certain systems vulnerable. This integrated automation employs a series of probes and watchdogs that revert the configuration to a properly functioning state that is kept in the backend databases within configurable intervals.
Using the same example, changes to ACL can be deployed to hundreds, if not thousands of devices with the click of a button. What makes this even more powerful, however, is that an engineer can input an ACL rule in a single format once, and the automation platform can in turn transform this rule, so that it can be enabled on a variety of devices in a mixed-vendor environment. Thus, through Network Automation and ZTP, operators can eliminate vendor lock-in, increase adoption of new platforms, and reduce overall network costs.
This post is an except of the original article  published by Channel Vision in July-August edition written by Ian Reither and Jason Craft. The second part is now live and dives into  Strategic Peering.
Share on Social
linkedin logo

By using the site, you agree to our use of cookies. Accept and close Find out more here.