SignalFx exits stealth with $28.5M in backing, promises Facebook-caliber systems monitoring off the shelf
As a budding startup, it can be challenging to say the least to hire the best and brightest technical talent in this industry. Often that talent is locked inside giants like Facebook, Google, and Amazon or, otherwise is intent on setting out to chart their own course. But fortunately, much of the knowledge gained inside these industry leaders can now be acquired in the form of the products these ex-mothershippers build after they leave.
As of today, that’s very much the case for the systems management space, thanks to the launch of SignalFx, a monitoring system built by one of Facebook’s lead software architects, Phil Liu. Prior to Facebook, Liu was Ben Horowitz’s “guy” Opsware, according to a blog post today by the Andreessen Horowitz founder, leading a company-saving product re-architecture over a grueling 9 month sprint. Liu’s co-foudner and SignalFx’s CEO is former VMware VP Products and Marketing Karthik Rau. Collectively, they’ve encountered and helped solve more systems management issues than most companies will see in a lifetime.
SignalFx (fka, SignalFuse) exited stealth today, announcing $20 million in Series B funding led by Charles River Ventures (CRV) with participation from existing investor Andreessen Horowitz* (A16Z). The round brings SignalFx’s total funding to $28.5 million. CRV general partner Devdutt Yellurkar will join Horowitz on the SignalFx board.
Ironically, SignalFx is addressing many of the same challenges as Opsware sought to, but is doing so in an entirely different era. Today, distributed architectures, cloud applications, and mobile adoption demand new monitoring solutions. Facebook, for example, revealed several years ago that its internal configuration and monitoring tools was processing more than 3 trillion metrics per day, a figure that has surely increase since then.
SignalFx’s secret sauce is called Signal Flow, a streaming analytics technology that allows companies to process massive volumes of metric data in real-time, and deliver near-instant analytics – something that required 15 minutes to one hour with prior-generation solutions. When you’re monitoring things like credit card fraud, the difference between right now and in 30 minutes can be the difference between actionable data and too late.
“There has been a fundamental shift in architecture, with people moving to new modern cloud architectures, all of which are microservices-based and distributed,” says CRV’s Yellurkar, who is also on the boards of Zendesk and OneLogin. “They all need scalability and performance and availability all the time.”
“Modern monitoring is now an analytics problem,” Rau says. “We’ve built for scale.”
Facebook and the rest of the top 1 percent of companies in the industry have the resources to build proprietary solutions internally to deliver these kinds of results. But for the rest of the world, modern system monitoring was never possible off the shelf until now. The closest many have gotten is through cobbling together existing open source solutions with only modest success. SignalFx, on the other hand, is flexible, meaning that it can be implemented across a universe of different systems architectures, and offers a public API that third-party developers can build on top of.
“We have a lot of insight into the problem space, but we ended up building a very different solution here [than at Facebook],” Liu says. “We had the luxury of building from the ground up using modern open source and proprietary technologies.”
Rau and Liu raised SignalFx’s A round of funding in March 2013 and quietly put the first beta version of its product in customers’ hands in Summer 2014. Today, the product is available as a self-service free trial, and will be priced on a tiered subscription basis according to the volume of data processed. The most basic, aka “Bootstrap” tier of 5,000 data points processed per minute costs $75 per month. The largest non-custom tier, or “Enterprise” package covers up to 100,000 data points per minute and costs $1,400 per month.
SignalFx counts Yelp, Tapjoy, Syphony Commerce, and Chairish among its customers, as well as a number of notable enterprises that declined to be mentioned publicly, according to SignalFx’s founders. The company’s largest customers are processing tens of billions of data points per day, Rau adds. The primary challenge for SignalFx at this stage is to create market awareness of its solution and build out its distribution model. The good news is the company is solving a near-universal problem – the bad, that no one has productized a solution for modern architectures, so CTOs and system architects are unaccustomed to shopping for them.
“We back great founders, and that is obviously the case here,”Yellurkar says. “They’ve got the first part of the equation, the product, sorted out. Now they need to go out and build the sales and marketing infrastructure to support it. We tend to get very involved in helping companies scale and plan to do the same here.”
As SignalFx enters the market with this first of its kind solution, it is inevitable that the company will see fast followers. This is simply too big of a market opportunity for there not to be. CRV and A16Z are betting big that the SignalFx has a sufficient head start and enough pedigree in its founding team to win what will surely be a tens-of billion dollar space.
Rau and Liu have built an engineer-heavy team of more than 20 people, including additional veterans of Facebook, Google, Yahoo, Okta, and Quora. By year’s end, the founders expect that headcount to swell to at least double its current level. The San Mateo-based company is currently operating out of YouTube’s old office space, a fact that is sure to lend some good vibes, not to mention systems architecture war stores to its cause.
Recalling the dark days rebuilding Opsware, Horowitz writes:
We had done it. It was a new day and we weren’t going to squander it. I will never forget that moment with Phil. When things go horribly wrong, the greatest people distinguish themselves. Phil could have made many excuses and blamed many people — most of all me. But, he would have none of that. Instead, he simply found his greatness.Sounds a lot like the kind of guy you want running your systems architecture. Unfortunately, Liu isn’t looking for a job. The closest you’ll get is buying his software, which is on store shelves today.
[*Disclosure: Andreessen Horowitz partners Marc Andreessen, Jeff Jordan, and Chris Dixon are investors in Pando.]
[Disclosure: Michael Carney has accepted a position as an associate at Upfront Ventures that begins in April. To the best of Pando’s knowledge, the companies in this post and their competitors have no affiliation with Upfront. This post went through Pando’s usual editorial process.]