Published
10/01/20
written-by
share

In this episode of GV News, Kelsey Hightower and I speak with Pixie Labs founders Zain Asgar and Ishan Mukherjee. Pixie Labs launches today to give developers a powerful toolset to monitor the performance of applications running on Kubernetes. Ishan and Zain cover the challenges in understanding and debugging distributed systems, and we all geek out over the latest in the Kubernetes ecosystem.

I've known Ishan for more than a decade. We first met when I was on the executive team at Kiva Systems, an enterprise robotics and software platform that was eventually acquired by Amazon to automate its warehouses. Ishan was a talented young engineer, fresh out of MIT with a lot of ideas about how the company could solve more of our customers' challenges.

Part of my job at Kiva was to lead customer-facing organizations. If something went wrong with one of our multimillion-dollar robotic systems in the field, I got the call. (These calls invariably seemed to come as we were all heading home late in the evenings.) Several of our teams would end up spending much of the night trying to recreate, identify, and fix the problem. Many times, the cause was human error — people were interacting with the system in ways that we hadn't anticipated or understood. Such experiences helped crystallize a few important things: 1) like many companies, we had inadvertently outsourced part of the debugging feedback loop to our customers, 2) customer empathy is at the core of designing great products, and 3) acquiring product feedback before the customer encounters an issue would be far preferable to waiting until the customer notices.

Ishan went on to lead Kiva's data science team, where he built close ties with customers and spent most of his time embedded at customer sites deeply familiarizing himself with their business objectives from the ground up. He has built his career around this customer-centric philosophy, and over the years we've worked together on two other GV companies: Premise Data, and Lattice Data (acquired by Apple).

Two years ago, he came to me with a new idea: give developers deeper visibility into application performance. His approach would harness autotelemetry to instantly troubleshoot applications running on Kubernetes. Ishan's inspiration was rooted in our early days at Kiva: customers and professional services teams shouldn't be the only ones with direct visibility into product performance - developers would have the tools to monitor their applications directly and thoroughly.

Ishan teamed up with Zain, an EIR at Benchmark who was previously part of Jeff Dean's Machine Intelligence Group at Google and who is currently an Adjunct Professor of Computer Science at Stanford University. Earning his PhD in Electrical Engineering from Stanford, Zain was employee #6 at Trifacta and prior to that, a senior engineer at NVIDIA. GV helped connect the founders to key early customers and a key Pixie advisor in Kelsey Hightower, a leading voice in the developer ecosystem and expert at the intersection of both customer empathy and Kubernetes. Together they have been working hard to improve performance monitoring for apps running on Kubernetes. They introduced their beta in May, and have seen strong validation from early customers signing six-figure annual contracts spanning multiple years.

As Kelsey notes, "With Pixie, there's no more reverse engineering what your customers are going through. Developers can see all the data about their infrastructure, all the time. You can do active debugging in a data-driven way."

I'm proud to see this team emerge from stealth today with a set of tools that will, in Kelsey's words, "make the hard things easy" for developers.