Seven Questions for Splunk CEO Godfrey Sullivan
You can’t get very far in any discussion about enterprise technology these days without hearing about “big data.” That’s the notion that there is useful business intelligence to be gleaned from combing through the huge volume of data generated by day-to-day operations in order to learn patterns that tell you things you wish you had known a lot sooner. Measuring customer buying patterns, traffic flows on a highway system and supply chain data are all examples of big-data problems.
Usually the phrase refers to the kind of heavy-lifting computing that involves big databases and specialized software to sort it all out — usually sold by the likes of large, serious software or hardware companies like Oracle or IBM.
But there’s another kind of data worth sifting and sorting, and while strictly speaking it may not seem as big at first, it’s turning out to be equally sizeable. It turns out there’s a lot of useful data being generated by machines of every kind. When I say machines, I mean it in a fairly broad definition: Web servers, cellphone towers, air conditioners. If a machine does something that repeats many times a day, hour or minute, then it is generating data that can be measured. And if it can be measured, there’s a pretty good chance you can do so using Splunk, software put out by the privately held company of the same name. It’s aimed at keeping minute-by-minute track of almost any kind of machine and turning that information into useful data that might help solve a problem before it starts, or yield useful information that improves a business.
Backed by venture capital investments from August Capital, Sevin Rosen Funds, JK&B Capital and Ignition Partners, Splunk is rare among private start-ups in that it discloses its annual financials: In 2010 it reported $66 million in sales, which amounted to 96 percent growth over the prior year. And this year it has grown about 70 percent again, which would put it on track to break the $100 million mark. Chatter about a possible Splunk IPO is already starting to bubble. CEO Godfrey Sullivan is a veteran of running public software companies, having helmed Hyperion Solutions from 2001 until its $3 billion acquisition by Oracle in 2007. I asked Sullivan about Splunk’s IPO intentions and many other things in a conversation earlier this month.
AllThingsD: At a high level, what is Splunk and what does it do?
Godfrey Sullivan: We have a proprietary database. Our core intellectual property is the ability to harvest all that data that gets generated by a machine. Splunk sucks it all in at a very high speed. We also have a user interface on top of it that looks a lot like Google. You can do command-line scripting and search for a particular combination of events, or you can build your own dashboards and say anytime some set of conditions reach a certain point, you can generate an alert.
Who uses Splunk and what do they do with it?
Our biggest user is probably Salesforce.com, but we’re also seeing a lot of use from Zynga, MetroPCS, T-Mobile. They index more than a terabyte a day. Zynga uses Splunk to monitor the performance of all its games — FarmVille, Mafia Wars and the like — all in real time. At first they started using it to identify problems, but now they use it for quality assurance in their development and test environment. All the developers at Salesforce use it for quality assurance, and to monitor performance and user behavior on Chatter.
We’re also in 150 government agencies, mostly for security. Agencies use Splunk to watch for attacks coming from bad IP addresses, and then watching to see if any employees have a history of touching those IP addresses.
You’ve also seen your use cases grow outside of Web servers recently, correct?
We’ve seen an explosion of use cases, and we’re trying to respond to that. Our user interface is one that IT administrators really love, because you can search on it or create reports very easily. But this explosion of use cases is people who don’t want to do that. They want it to manage windmills or smart buildings. We’re starting to get a lot more brand awareness of Splunk as the way to manage machine data. It could be knowing where a box of wine is somewhere out on a UPS truck, or it could be pulling call-history data off a cellphone, and so we’re really trying to respond to this big explosion in new use cases.
Have there been any examples that have surprised you?
There was a guy in London who said he was managing a greenhouse with it, taking temperature readings and driving levers to open and shut. There was another one in Japan, just after the tsunami. The local aid agency there was using Splunk to help distribute food. They were pulling weather data and information about which roads were open or impassable, to understand where to send food and in what direction on any given day, and to keep it from spoiling en route. There was another we heard about in China that was for tracking husbands and wives from cellphone data. We got called in to appear on a TV talk show about that once.
Do you have anything new teed up for the remainder of the year?
We just announced our first cloud service. We’ve always been run in the cloud, because our customers would install it to monitor an app, and their infrastructure was in the cloud. But for the first time we’ve started offering Splunk as a service. We call it Splunk Storm. The essence is that it’s aimed at developers in the cloud. If you’re a developer, you can go out and turn on a Splunk instance, and set the dials for how much indexing and data storage you want. And once you’ve done that, you can Splunk all of your development work as you do it in the cloud.
Splunk is unusual in that it is privately held and venture-backed, but you report some of your financials. Why is that?
We publish numbers once a year, and we did $66 million in revenue last year. And we just announced that in our second quarter we grew 70 percent over the year-ago quarter, and that comes on top of 68 percent growth in the first quarter. So life is good. We have 3,000 customers in 70 countries.
These days, hitting $100 million in sales is considered sort of a magic number for going public. If you do that — and it looks like you will — will you consider going public?
I am no longer allowed to answer that question. I will tell you we’ve recently hired a CFO, a general counsel and a CIO. Beyond that, I really can’t answer that question.