This week I joined Augtera Networks as Sr. Director of Product and Product Marketing. I’m not one to jump ship lightly – my tenures at Kentik and Arbor Networks were 7 and 11 years respectively. So transitions like these create an opportunity for me to stop for a moment and reflect on the industry’s past and future, and my own as well.
I’ve spent my entire career in the world of infrastructure. Much of it relating to networks, but also systems and software. For as long as I’ve been immersed in it, the state of infrastructure has always been one step ahead of our ability to operate it as effectively as we’d like. I’m old enough to remember when it seemed impossible just to generate the data we needed to properly troubleshoot networks. The days when our best tooling was a packet sniffer on a cart in the corner of the data center.
Of course, generating infrastructure telemetry is no longer the problem. Over the last 20 years, technology has progressed to overcome many challenges in dealing with large volumes of data – generating, transporting, storing and processing it. In fact, most organizations are now drowning in infrastructure data and really struggling to make sense of it. Reactively, yes – when there’s a known problem, we now have software architectures that allow us to dig through huge data sets to quickly understand the root cause in many cases. But proactive use of infra data, to detect problems as they happen – or even precursor signals before failures occur – was still a largely unsolved problem.
Enter AI / ML. Until recently, I would have said that these emerging technologies were more hype than reality, with practical applications limited to niche use cases at industry giants. That sentiment came partly from the industry’s many failed attempts to apply AI / ML to infrastructure data. With products that seemed like mixing some data and ML algorithms in a blender and hoping for an AIOps smoothie.
Innovation is funny though, built on top of a mountain of failures. Once-impossible inventions move ever closer to the horizon of possibility. The sunrise of innovation isn’t as visible as Sol rising in the East however, and it’s often a team of domain experts who see it first – the moment when progress in one discipline can be applied to intractable problems in another discipline. This is the moment when start-up magic can happen, and what really attracted me to Augtera. A team with deep expertise in both AI / ML and networks, collaborating to solve the thorniest problems in network and infrastructure operations.
While it was the team that sparked my interest, it’s what they’ve already delivered that sealed the deal. Augtera’s platform is truly finding needles in haystacks in large global networks. Not just real-time detection, but entire categories of failures that weren’t anticipated by customers. Unknown unknowns – like distributed temperature increases across devices predicting an HVAC unit that failed a few days later, or finding misconfigured SLA policies because they were causing SDWAN traffic redirections that didn’t correspond to underlying network conditions. More than 50% of the anomalies Augtera detects result in customer remediation actions. Anyone who’s dipped their toe in network monitoring knows what an amazing stat that is. Most alert streams contain so few actionable events that they’re quickly ignored.
I’m excited to join the team at Augtera who have finally moved the state of the art in operations ahead of the state of the network. And we’re just getting started.