Déjà Vu All Over Again with Forward Networks

It is déjà vu all over again for me with Forward Networks. I was fortunate to join Bob Metcalfe in early 1981 to help build 3Com, which became one of the early pioneers that launched the concept of computer networking. While 3Com was most known for its Ethernet products, a little known fact was that our first product was UNET, Unix software for networking. UNET was the first commercial version of Unix with integrated TCP/IP protocols and a software driver for Ethernet connections. Our first customer was a little known company in Seattle named Microsoft who needed to network their three Digital Equipment (DEC) minicomputers to accelerate the development of a secret software project later to be known as DOS, the operating system for IBM PCs.

Maintaining reliable operation of Microsoft’s three computer network was quite simple: all you had to do was make sure the right LEDs were blinking in the right order. Now some 35 years later, data center networks consist of several thousand network devices and tens of thousands compute servers. These large networks have resulted in several orders of magnitude in the complexity of assuring reliable operation. This leads to why I am excited to introduce you to Forward Networks, whom are creating a new category of software, named Network Assurance, to solve this problem.

Why did a16z decide to become the lead investor in Forward Networks? Three reasons (oops there I go again doing things in 3s): the team, the technology, and the market opportunity.

How often do you get an opportunity to invest in a team of four PhD computer science graduates from the research lab of Nick McKeown at Stanford, who is recognized as one of the early pioneers in the growing market of Software Defined Networking. And this team of exceptional networking software researchers had proven their ability to work together over a period of 5+ years at Stanford, and an additional year at Forward Networks, creating the foundational technology behind the Forward Platform. They wowed me and my colleagues at a16z when they demonstrated their working prototype during their investment presentation.

Simply put, the technology behind the Forward Platform is unique in its features and functionality, and essential to the reliable operation of our increasingly complex computer networks. The fundamental underlying technology is a series of revolutionary algorithms that create a mathematical model of an entire functioning network. By creating, in software, an accurate copy of the network, Forward Networks gives network operators the ability to easily visualize and search complex networks, quickly debug problems, verify policy correctness at scale, and predict network behavior prior to making changes to production equipment.

What about the market opportunity? What problems does the Forward Platform solve? Recall the last time you could not access the network, or it had slowed down to a crawl. You were beside yourself. We are addicted to our network being always on and ready to respond, whenever we need it.

However, somewhere in the world, thousands of people experience a network outage every day. Think about the massive Comcast or Southwest Airlines outages among others. On average, large enterprises experience at least five outages per month at a cost of $740K for each outage. This occurs in spite of these enterprises spending tens of billions of dollars per year on people to staff and operate their networks. And to add insult to injury, over 80% of the network outages are caused by human error. Imagine the dollars saved, and the happiness of your customers if you could make significant improvements to the reliability of your network’s operation.

Why is it this way? Given all of the money we spend to staff and operate our networks, why is that humans are constantly breaking them? There are three (oops there goes doing things in 3s again) key aspects about the challenges facing network operators today:

  1. Today’s networks are both extremely large and highly diverse. Large enterprise networks range in size from tens of thousands to hundreds of thousands of network devices coming from a variety of network equipment vendors. Then, each network device could have from hundreds to hundreds of thousands of rules to properly operate them. Massive complexity.
  1. The tools that are available today to aid network engineers in operating their networks are woefully inadequate. Most are some 30 years old, going back to my days at 3Com. They were developed for networks that were much simpler, like that first network of three minicomputers we connected at Microsoft with UNET and Ethernet. Development of new tools to improve operations of networks has been somewhat of a backwater over the intervening years, until now with the introduction of the Forward Platform.
  1. Thousands of network devices with thousands of rules and you are only one network rule misconfiguration away from a massive network outage, despite having complete hardware redundancy.

Extremely large and highly diverse large networks that are being managed by tools 30 years old, and that are one human misconfiguration away from a business stopping outage, sounds like a large market opportunity in need of creative new solutions.

The Forward Platform enables applications that provide network operator-assisted search, operator-assisted verification, and operator-assisted predictability resulting in making the operation of networks much safer and reliable. In this new, emerging category of networking software called Network Assurance, I expect we’ll see, in the not distant future, autonomous, self-operating networks that virtually eliminate network outages, giving us the always on, always available network that we desire.