I recently had a huge revelation about networking. You see, I’ve spent almost 25 years working in IT, specifically data communications (yes I’m that old!). I had run into so many incredible products that never succeeded in the market, and I was sitting there, beverage in hand, thinking “Why do so many good products fail to be adopted?” It almost seems like the best products with amazing new technologies fail more often than the boring ones.
The simple answer is TIMING. Timing is everything. I’ve worked for companies that were building incredibly cool solutions. One of them was doing something strange about 8 years ago called “Zero Trust”. No one seemed to care about the tech, and the term was not resonating. Today there are probably a hundred ZT or ZTNA startups trying to be the next unicorn. Timing is everything.
Sometimes there are startups productizing incredibly obvious bits of technology that everyone agrees upon, but they still fail to make a change in the market. For example, everyone now agrees that using linux as the OS for a router or switch makes sense, but 10 years ago this idea was somewhat bizarre. Today you’d be hard pressed to find a NOS that isn’t based on linux, or at a bare minimum offer a linux “guestshell” so you can run your automation scripts in a familiar platform.
Changes unto themselves are often not worth the effort. The cost of changing something to lower your purchase costs can often mean a big change in how your network operations team manages the gear, it’s likely not worth the headache so you typically keep doing it the same way you’ve always done it. But what about when the entire platform changes, not just in your company but for the entire world?
Platform Transitions Matter
Platform transitions usually happen when there’s no way people can live their lives the way they used to, or when there is a changing of the guard. As old people die and retire, new people refuse to do it the old way, they throw out the old runbook, or they automate around the repetitive bits.
Putting Linux on a network switch is a good idea but by itself it was not enough to cause network architects to upgrade their devices and change their operational models. This new way of programming networks was not attached to a major market transition, and as a result it was not compelling enough for end users to migrate.
With that being said, why in the world am I so excited about ANOTHER open network operating system? I must be a glutton for punishment, right?
There are two major things happening in networking today.
SONiC is here and it’s a game changing platform transition
SONiC has huge momentum in the market. Network architects need to be thinking about how to ride the wave, standing up straight and letting it hit you is a sure way to fail. There is a major market transition about to take place. What does SONiC offer?
- A standardized way of programming network devices
- An open and composable operating system that can be tailored for each use case
- Supply chain freedom – you have hundreds of options including favorites such as Cisco, Arista, Juniper, Nvidia, Dell, and more
- Open Source always wins in the long run
But managing SONiC has not been a walk in the park. This isn’t your grandpa’s CLI router. And who manages a single device anyhow? Today’s networks are carefully constructed distributed applications (sometimes called fabrics) because single points of failure are a sure way to force you to update your resume.
If only there was a great open source technology for managing a bunch of devices running containers…
It’s Kubernetes silly!
It should be obvious to everyone reading this that Kubernetes is the de-facto technology for application orchestration. And really, Kubernetes is the primary method used to control YOUR infrastructure. But why has the network lagged so far behind in this new Cloud Native world we’re living in? There are lots of GREAT Container Network Interfaces (CNIs) and Kubernetes Networking Providers, but those really only address what happens once the application traffic has reached the ingress for your cluster. You still have to plug all these k8s nodes into something.
Two platform transitions are better than one! What happens when these two waves merge?
If you’re like us, you know the answer is an incredible new way of building and operating your equipment. But don’t take our word for it, just look at the cloud. How often do you configure the physical network for your Kubernetes clusters in the cloud? NEVER. You don’t even think about it. And application owners get fat, dumb, and happy when they don’t have to think about the network. But try running Kubernetes in your own data center or edge, not so fast! All of the sudden you have to deal with the legacy network operational model again, and this is a great reason why people will stay in the cloud and pay 5-10x to host their workloads. Because it’s been too difficult.
But not anymore
At Hedgehog, we are fusing the native Kubernetes control plane to the open source networking goodness that is the DNA of the SONiC community. And can you tell I’m excited about it?
Through a series of blog posts, I’m going to be revealing the answers to the questions that are probably running through your head right now. I promise you, this is going to be very exciting, and like any good presenter, at the end I’ll be giving you candy (in the form of free open source software and a turn-key onboarding model).
So why should you evaluate Hedgehog?
- The same API and operations that you’re using for the rest of your infrastructure can be used to provision and manage your physical network.
- Finding people who can manage k8s and know how to run applications in it is much easier than finding people who can manage multiple vendor proprietary NOSs.
- You can consider all of the available hardware vendors, improving supply chain options and minimizing business risk
We’re hard at work on building the best product, networking so simple that it seems to disappear once you turn it on.
P.S. If you’d like to hear more about our revolution, please request a meeting and we can answer all of your questions.
Josh Saul has worked on advancing the state of the art in data center networking for more than 25 years. As an architect, he built core networks for GE, Pfizer and NBC Universal. After jumping to the vendor side at Cisco, Josh focused on the Fortune 100 financial sector and evangelized new technologies to customers. Building on technical successes, Josh led marketing and product teams at VMware (acquired by Broadcom, Cumulus Networks (acquired by Nvidia), Apstra (acquired by Juniper), and Netris. Josh lives in New York City with his two children.