• Human Infrastructure
  • Posts
  • Human Infrastructure 389: The Difference Between Network Automation and Automating Your Network

Human Infrastructure 389: The Difference Between Network Automation and Automating Your Network

AuthorAuthor

Ethan Banks & Drew Conry-Murray
February 20, 2025

THIS WEEK’S MUST-READ BLOGS 🤓

Mark Prosser has a thoughtful piece that draws a distinction between “network automation” and “automating your network.” So what’s the difference? You should read the whole thing to get Mark’s perspective, but at the risk of over-simplification, Mark is drawing a line between one-off automations that handle discreet tasks vs a set of processes that operate across many systems and rely on one or more data stores, workflows, signals, and other inputs from multiple sources.  

Mark’s thinking draws on ideas from Claudia De Luna, Dinesh Dutt, and others in the network automation world, which Mark acknowledges. He borrows this example from Claudia: Is writing and using a script that checks a route on a switch network automation? Yes. But is that script automating your network? No.

While this distinction may seem like a semantics exercise, I think Mark and others are describing a way of thinking about automation that aims to move the industry beyond one-off jobs to something much more comprehensive and valuable. - Drew 

Brad Casemore comments on Juniper CEO Rami Rahim’s observations that the US Department of Justice is getting it wrong in their suit to block the JuniHPEr merger. I haven’t (thus far) seen any analyst that agrees with the DOJ’s position, as it seems untenable on the merits.

<speculation>Could it be that this is really about delaying the transaction? If so, who would benefit from such a delay?</speculation> - Ethan

This site is pretty much The Onion, but for tech nerds. Author M. P. Labelspace writes pieces with such headlines as…

Glad to see this site back to publishing new content after being dormant for a few years. We need more humor in the world! - Ethan

Jens is teasing you with this title. He doesn’t actually think v6 is hard. He thinks it’s different. He walks through Happy Eyeballs and its impact on client behavior, then does a little diagnosing of a reported problem with traceroute -T. He concludes by observing that some people think IPv6 is hard because the v6 resource they’re trying to hit wasn’t set up correctly. “So please: If you do IPv6 take it seriously. If you don’t take it seriously, don’t do IPv6. That leaves to people thinking that IPv6 is hard and can not be done.” He has a point. - Ethan

Nick Buraglio describes the different operating modes available for IPv6 environments as defined by RFC8925. Nick notes that these operating modes “are easy to understand definitions for most operating modes that an engineer will see in the wild. As this is an RFC that defines an operational option for running a network, it is very complete in its description of operational models.”

The modes include IPv6 only, IPv6-mostly, dual-stack, and others. And BTW, Nick is the new co-host of the IPv6 Buzz podcast on the Packet Pushers network. - Drew

Jeff McLaughlin isn’t all aboard the AI hype train. He brings a reasonable skepticism to the current capabilities of AI tools while also experimenting with them to learn where they can be useful. For instance, he’s found ChatGPT to be sometimes helpful. However, he notes “if we want to go from useful tool to fully agentic network management, well, we have a long way to go. Right now the agentic demos feel a bit like the Alexa ones–the art of the possible, but not necessarily the probable.” - Drew 

MORE BLOGS

AutoCon3 Early Bird Registration is Open!
Early-bird registration for the AutoCon3 conference is now open. Join us in Prague from May 26 to the 30th for the industry’s premier network automation event! AutoCon brings you hands-on workshops, deep-dive presentations from engineers deploying network automation in production, and hallway-track conversations with your peers. Whether you’re just starting your network automation journey or pushing its boundaries, AutoCon is THE place to gather, learn, and share.
SEE EARLY BIRD PRICING AND REGISTER HERE

TECH NEWS 📣

The headline alludes to Uncle Mark’s Project Waterworth, a 31,000 mile (50,000 kilometer) subsea cable he’ll use to deliver algorithmically-driven content to regions as diverse as the United States, India, South Africa, Brazil, and more. 24 fiber pairs. The swipe heard ‘round the world! - Ethan

Introducing Very Low Earth Orbit (VLEO) satellites. Should they come to fruition, VLEO satellites will fly at the top of Earth’s atmosphere, the “sweet spot” expected to be between 125-155 miles (200-250km). How will they overcome drag and stay aloft? That’s the problem research is trying to solve.

“A handful of pioneering companies have begun work on designs for satellites that may be able to orbit the planet at these unusually low altitudes while simultaneously harvesting air and using it to make propellant – literally on the fly. … Such a system, called Air-Breathing Electric Propulsion (ABEP), has been made possible by advancements in electric and ion propulsion in recent years. In essence, it involves fixing a large bucket or opening to the front of the satellite, into which gas molecules from the atmosphere flow before they are ionised to create plasma that generates thrust.”

Companies Kreios Space hopes to launch a demo VLEO satellite by 2026 and Stellar Advanced Concepts by 2027. Other companies are in the mix as well. - Ethan

MORE NEWS

FOR THE LULZ 🤣

RESEARCH & RESOURCES 📒

Here’s the abstract. “Networking stacks are the backbone of communication and information exchange. This paper investigates the TCP/IPv4 and UDP/IPv4 network stack of Linux, the most common server OS. We describe a trace of the most critical networking functions of the Linux kernel 5.10.8. Although Linux networking code documentation exists, it is often outdated or only covers specific aspects like the IP or TCP layer. We address this holistically, covering a packet’s egress and ingress path through the Linux networking stack. Moreover, we highlight intricacies of the implementation and present how the Linux kernel realizes networking protocols. Our paper can serve as a basis for performance optimizations, security analysis, network observability, or debugging.” - Ethan

From the README. “Vulnerability-Lookup facilitates quick correlation of vulnerabilities from various sources, independent of vulnerability IDs, and streamlines the management of Coordinated Vulnerability Disclosure (CVD). Vulnerability-Lookup is also a collaborative platform where users can comment on security advisories and create bundles.”

Coordinates with several advisory services, databases, and feeds you’ve heard of, offers an API, and is written mostly in Python. - Ethan

There are other tools with this name, so to disambiguate, here’s an excerpt from the README. “Nping is a Ping tool developed in Rust using the ICMP protocol. It supports concurrent Ping for multiple addresses, visual chart display, real-time data updates, and other features.”

Image of Nping in action taken from Github repo.

MORE RESOURCES

  1. kwatch (detect K8s cluster crashes) - abahmed via GitHub

  2. ESPARGOS: ESP32-based WiFi sensing array - espargos.net

  3. d2c.sh (Dynamic IPv4/6 records for Cloudflare) - ddries via GitHub

Meet the Packet Pushers at DCD Connect!
Join Ethan Banks and Drew Conry-Murray at DCD Connect in NYC this March! DCD Connect brings together more than 3,500 senior leaders working on some of the largest data center projects in North America. Register here and use the code PACKET and we’ll see you in New York! Ethan and Drew will be on site March 24th. Come say hello and collect your free virtual donut!

INDUSTRY BLOGS & VENDOR ANNOUNCEMENTS 💬 

Bandwidth consumption trends indicate that we’ve got enough pipe coming off the towers to our mobile devices. The problem 6G should be solving isn’t more bandwidth, as most mobile applications need 15-20Mbps or less (that includes 4K video according to Netflix), and an app needing more than 1Gbps is hard to find. This is why that even though 5G can provide significantly more bandwidth than 4G, the experience doesn’t feel any different.

What if we look to the future, speculating about possible bandwidth requirements for the metaverse, AI, and stereo holograms? Even there, it’s hard to see demand beyond 1Gbps, which is within 5G’s capabilities. Thus, it’s hard to argue that 6G needs to be faster for purposes of future proofing mobile network infrastructure.

If all this remains true as the next few years pass, then what’s the motivation for telcos to invest in 6G infrastructure? Why would people upgrade to 6G if 5G isn’t a constraint, and 5G to 6G (like 4G to 5G before it) offers little to no perceivable improvement? Seems like a big outlay without a timely ROI. - Ethan

The Infrahub team reports, “Emma is the experimental little sister of Otto, and she is to help you interact with Infrahub. Currently, Emma can help you to:

  • Import CSV Data into Infrahub

  • Export Data from Infrahub in CSV format

  • Build, Load, and Visualize the Infrahub schema

WARNING: Emma is experimental by design, and may not work as expected. She tests out new ideas before her big brother Otto gets involved and implements them in Infrahub.” - Ethan

Alkira is extending its Network Infrastructure as a Service platform with a new load-balancer offering. From the press release: “By deploying load balancers directly into Alkira's Cloud Exchange Points (CXPs), the service provides optimal placement and connectivity that eliminates complex networking configurations. This native integration enables organizations to deploy load balancing services with a simple point-and-click interface while maintaining full access to advanced load balancing capabilities.” Alkira is launching with F5’s Big-IP, but says other choices will follow.  - Drew 

This post from Dell’Oro analyst Mauricio Sanchez provides an overview of, and some larger market context for, Palo Alto Networks’ CNAPP offering. What is CNAPP? It stands for Cloud-Native Application Protection Platform. It’s a market category that aims to integrate a variety of cloud security technologies that address risks across the lifecycle of a cloud application. Sanchez notes that the CNAPP market is currently worth $700 million. And while Palo Alto Networks is the leader, rivals such as Crowdstrike and Wiz are enjoying dramatic growth. If you’re looking for some context and overview of the CNAPP market, this post is an excellent start. - Drew  

If you’re on the Aviatrix platform, the company is rolling out a new course in its Aviatrix-specific certification program. The Hybrid Cloud Speciality course covers cloud networking issues, including:

  • Secure and high-performance hybrid cloud using Aviatrix Edge for datacenter/colocation

  • Secure and high-performance hybrid cloud using Aviatrix Edge for Equinix

  • Secure and high-performance hybrid cloud using Aviatrix Edge for Megaport

  • Secure hybrid cloud using third-party devices in on-prem data centers and collocations

The first instantiation of the course takes place March 4th. You can register for it here. - Drew

The post provides a good overview of the benefits of Multi-Link Operation (MLO), one of the key features in the Wi-Fi 7 standard that improves performance over previous versions. In short, MLO lets APs and wireless devices communicate over multiple bands simultaneously, which can improve throughput, reduce latency, and take full advantage of available bands. The post then provides an illustrated and detailed example of how MLO’s STR mode works (using Cisco Catalyst gear). - Drew 

Versa has announced the general release of its Sovereign SASE offering in which Versa’s security services can be run on the customer’s own infrastructure. Versa says its Sovereign SASE offering is “Delivered via dedicated gateways in customers’ infrastructure under customer management and control – completely air-gapped.” 

To me, that sounds less like SASE and more like a UTM appliance, or virtualized security services running on customer hardware on prem. And given that the headline says Versa is “redefining SASE,” maybe Versa thinks so too? Pedantry aside, if you’re an organization facing strict regulations about data sovereignty or are grappling with other constraints that put traditional SASE out of reach, Versa would like a word. - Drew 

 

MORE INDUSTRY NOISES

DYSTOPIA IRL 🐙

TOO MANY LINKS WOULD NEVER BE ENOUGH 🐳

LAST LAUGH 😆

This is from a data scientist, but I think it tracks pretty well for the Human Infrastructure crowd. Shared on Bluesky by @cghlewis.bsky.social