Episode 10: Interview with Rob Das, Chief Architect & Co-Founder at Splunk

September 28, 2017


Pramod Shashidhara

Rob Das co-founded Splunk and has served as Chief Architect since 2005. He also served as Vice President, member of the board of directors. Previously, Mr. Das has been a large-scale, distributed software architect and engineer for both early-stage ventures and large companies, including Avolent Inc., an application software provider, Lotus Development, a software company, and Sun Microsystems, Inc., a computer systems company. Mr. Das studied computer science at Indiana University.


Pramod: Hello everyone. Welcome to another episode of Mapping the journey. Since I started this podcast I always wanted to speak to the founder of a multinational successful company founded here in Bay Area. You know hundreds of start-ups start and fail every year. I’m curious to know what these successful companies did differently and how hard it is to build one. Today I have got this opportunity to speak to the co-founder of Splunk Mr. Rob Das. He’s an entrepreneur currently serving as a chief architect.

For all the listeners Splunk is a software for searching, monitoring and analyzing machine-generated data. We’ll hear more about the company from the founder himself. Without further delay, I welcome Mr. Rob Das to the show.

Rob: Thank you. Thank you very much.

Pramod: Rob we know you as an entrepreneur co-founder of Splunk. Tell me more about yourself.

Rob: Sure. So, I’ve been in the software industry for probably 35 years. Although since about 2012, I sort of stopped doing software. After working that long and that much-complicated stuff, it just you get tired after a while and you want to do something a little different. I was very very fortunate that’s Splunk turned out so well so I was able to actually retire in a beautiful wine country and do something completely different with my life so, it’s been really good. (Awesome),

But doing software for a really long time, as a matter of fact, I think one of the things that is really interesting is, I actually learnt how to program when I was in sixth grade. So that would be 1969. So, which basically tells you how old I am.

So my father, was a professor at the University of California in Berkeley and so I had access to Lawrence Hall Science and some of their labs and things like that, and I went to a camp actually for students, for children of professors at the University and actually got into learning how to program at a really young age, a very very long time ago and that stuck with me the whole time, I didn’t just do it and drop it. I actually stuck with that all the way through high school, all the way through college, all the way through a job, all the way up until I stopped and started doing something else with wine and grapes.

Pramod: Awesome…! so you had any other interests growing as a kid or just programming and things?

Rob: Well I mean, when I was young, I did a lot of sports like all kids of my age but one of the things is that I used to go, when we did this, it was on an HP 2000B which is a really old paper tape and teletype machines and learning BASIC. So, I took this class. My friend & I used to go on weekends instead of running around and playing with our friends, we actually went to this place and worked on programs like writing games together, stupid games like checkers or tic-tac-toe or whatever we could do. That was fun for us. We actually did that. But we also went outside playing tennis, some sports, some swimming… like all kids but I was probably one of the few kids that had the opportunity to actually do software that long ago in 1969 which was kind of cool.

Pramod: Yes awesome! So, you got a head start early. What made you take the leap into entrepreneurship. what was your first business idea?

Rob: Well so I have worked at larger companies but I found that I enjoyed the smaller groups of people much, in a much more dynamic environment and so after Taligent everything I had done since then have been start-ups and might be probably a total of seven or eight of them. So, tons of things, some mildly successful, others not, but I think you learn from both successes and your failures so it’s been a good, a good ride.

Pramod: So yeah you said you started seven companies. Like how do you get ideas?

Rob: Yeah so, to answer your question about how we come up with ideas, there’s a couple of different ways to do it. One is, I’d say that the way we’ve sort of done in the past is based on a problem that we’ve had in the past.

So, when we decided to do Splunk for instance, we literally said we’re gonna start a company. What do you want to do? Let’s think about this. Do we want to do a genetic database? Do you want to do this? Do we wanna do that? Well, eventually we said, well what were the pain points that we’ve had in our career.

So, for example managing groups of system administrators that are taking care of the equipment in the network in large data centers. What kind of problems do they have? What kind of problems did they have deploying software? I mean the answer is yeah, we’ve had common problems over a long time. How do we solve those? How did we use to solve those problems? We get in a conference room and we all meet and we would take a long time to solve it. There’s got to be a faster way of solving problems and through some sort of software and some sort of innovation. So, it’s all based on past experiences of problems that we’ve had and that’s sort of the way.

This is the way that we all go about it and then you validate that with customers, your potential customers. To make sure that it’s something that is a problem they have or we are the only ones that had that problem. The answer is yes, that makes sense to start going to tack it down.

Pramod: Ok awesome. So, you take the idea and you go talk to the customers and see whether it’s really a problem, that’s cool.

Rob: And we did that. And that it really depends on how much time you have, what kind of financial resources you have, can your family like take it? That sort of thing.

So, when we did Splunk, the one I’m just to give you an example here and I’ll try to do it quickly, what we found is that, when a transaction occurred in a networked environment like let’s say just for example you want to buy something on the e-commerce site and it touches all these different pieces of equipment and it runs through all these different software stacks and eventually that transaction comes back and says thank you for buying it. Everything is cool but if any one piece of software or any one piece of equipment on that path has a problem, it comes back and it couldn’t fulfill your request, that’s a bad experience for the customer. They don’t know what happened and this happens. And not just e-commerce transactions but every kind of transaction in the world has this problem and it touches all these different pieces and the more pieces there are to the puzzle, the greater the potential for a problem.

And so, what we found while Eric and I were working on a start-up company doing some sort of stupid transaction stuff, our biggest problem was whenever there was a problem, it would take us, literally it could take us days to find the problem. And where is the problem in this transaction flow? How do we find it? Well everybody’s an expert in a certain area and you have to get all the experts together in the room and you have to look at the code and you have to find out what it is and you look at the log files and you try to figure out what happened and when, and narrow it all down, Right?, and then you are like “Oh I found it” then you fix it. But that’s a very slow process, it takes a lot of energy and a lot of time. So, what we wanted to do was to somehow gather all those transactions up through that event data, log data so that we could help debug those kinds of distributed transaction problems and that’s kind of how Splunk got started.

So, we went and asked probably 60 different people, different companies, do you have this problem? How do you solve it? And ask them until we had it in our minds that this is a valid problem and this problem needs to be solved and the way to solve it is to use this log data in some way that makes sense.

We asked them a lot of questions not just “do you have this problem”, but “what’s your favorite tool?” A lot of them said search engines are their favorite tool, Google. What do you mean Google is your favorite tool? Well whenever we have a problem, we look it up on Google because somebody else probably has that problem. Well eventually, we kind of thought well wait a minute. Log files, Google! What if we put the log files in Google? Well, something like Google.

A lot of people search through the log files so we don’t have to get root on the machine, pull the log and take a look at it because the systems administrators did not want to give root access to the developers. And so, they would have to copy the file, ship them to the developers. Well, what if all that stuff was stored somewhere else and with your search engine, you can search through it and you didn’t have to get root access on any machines. Well everybody would be happy. That’s kind of how Splunk started.

Pramod: Nice. It’s hard to solve a problem without logs today. So, because the software size is so huge, so yeah, searching logs was a very good idea I think. Was this 2002- 2003?

Rob: 2002 we decided to start the company. and about eighteen months later or so, was when or maybe twelve months we started writing a little bit of code but it was different. it wasn’t quite Splunk. That was more transactions searchy stuff. Not really log file. I mean using log data but it was a very different type of thing but we ended up putting that together. I think we got funded in 2004 and very early 2005 we came out in public and said here’s like our alpha product or beta product. You take a look and we showed the people in Linux world.

Pramod: Okay. Awesome…! So, you had the idea like how easy or difficult was it to convince investors.

Rob: It was hard. Well, we were convincing them about transactions and transaction search and transaction failures and how to find them and all that. And all we had was a demo written in JavaScript and HTML and a PowerPoint presentation. So, no customers, just three entrepreneurs and like I said a PowerPoint presentation and a demo that really didn’t connect to anything or do anything. It’s hard when you don’t have that. Now, what was interesting is that Michael and myself and Eric had worked together before. And so, we were sort of a known quantity to a number of investors. So Michael had a lot of venture capital

Now, what was interesting is that Michael and myself and Eric had worked together before. And so, we were sort of a known quantity to a number of investors. So Michael had a lot of venture capital contacts. And we had made venture capital contacts through previous start-up companies. And so, we went through all those guys. So if you’re starting from scratch and you don’t have any contacts, it’s very hard to get funded unless you know somebody because everything is done through references, right? Anyway, we had the references so that’s good. We actually got to talk to the VCs which is better than most people can do. We went to 20-30 different VCs and many of them we went multiple times. it took a good year. I don’t know exactly but probably a good year to get funded.

And then we got our Series A from David Hornik at August Capital and that was half of it. He put in two and a half million and then we got two and a half million from Sevin Rosen Funds. So a total of five million for series A And then were incubated at Sevin Rosen, on the Emerson Street right across from the original Google office.

Pramod: So nice, you kind of persisted that one year. That is the most important period when you have the idea when you’re not finding the investors.

Rob: The hardest part is it’s really hard to focus on building something while you’re looking for money. I think that’s the hardest part. We were so consumed in getting our presentation right, and having our meetings and making sure that we got our point across and refining it from one pitch to the next. Like we would get feedback from someone we would incorporate that feedback back into the presentation back into the way we talked and we spoke. Then we do that again for the next one, we’d find out what their feedback was and eventually, it just got better and better and better.

More and more refined, we knew what to talk about when not to talk about and how to get an idea across quickly. How to get our elevator pitch down, all that sort of stuff it took a long time, and a lot of practice and a lot of rejection and a lot of feedback from people listen to our pitch and eventually we got it right. And I think what was interesting is that Eric and Michael and myself are, were a little bit older than most of the entrepreneurs that we’re looking for money at that point and I think that actually played to our advantage and the fact that we work together made a big difference and I think that Nick and David Hornik both thought that was a very positive thing.

Pramod: Awesome…! As you have already said if it wasn’t an overnight success. According to you what was the biggest hurdle in building Splunk? Yeah and then how did you overcome?

Rob: Uhhhh gosh, hiring people probably really hard. Hiring really really good people. And I think we did a tremendous job hiring people because the philosophy is, hire slow fire fast. Take your time hiring the people if the people don’t work out, let them go before they can become poisonous.

But we knew a number of people, we got started with them, we were comfortable with them. We also had the philosophy of hiring people that were smarter than ourselves. I think the right answer is if somebody knows more about you in a certain area you hire them, you do It, right?

Hurdles I mean gosh, there were technical hurdles, there were sales hurdles, there were all different kinds of them. I mean technically one of the more difficult things is this wasn’t running up to the cloud, that wasn’t really much by the way cloud infrastructure. So, everything was on-premise but everybody had different sort of hardware configuration and so we had this gigantic matrix of configurations and the whole concept of doing automated testing and putting together test suites so that we know we can run on different versions of the operating system on different hardware platforms and then porting. oh my god the technical, just that portion all along was incredibly hard. I think most of it is automated at this point. But just getting that all up and running was a tremendous amount of work. Absolutely tremendous amount.

On Selling the software-

Rob: One of the thing that was really interesting about what we did was, the way that we decided we were going to sell the software. And this was absolutely the smartest thing I think that we did as a company, is that in the past we had, we were selling enterprise software. Let’s say anywhere between a hundred thousand and half-a-million-dollar software, that’s a lot of money. (Yes), So, what that means is that when you sell software, you’re selling to the people that are very high up in the organization, sea-level people. And you’re out there on the golf course and you’re out there traveling around. And it takes a long time to close the deal.

And as a start-up company, they don’t take you terribly seriously because there’s some guy from IBM or CA, these are people they have relationships with, they don’t have a relationship with you yet so that they don’t want to buy it. But maybe they do buy it. That’s cool. So, we hire a staff of salespeople, maybe they’d sell one or two. And then it would go three months with no sales. And then maybe they’d sell one or two and then no sales. So, it’s kind of like a cycle, everybody’s happy. Oh…! we’re selling, yeah, we’re getting traction, we’re all gonna get rich. And then nothing. Oh god…! Now what I’m gonna do? So, you end up firing those salespeople. You hire another staff, the same thing happens. But that cycle’s hard and it really sucks.

So, we decided from day one, we were not going to sell our product that way. That is a terrible way to go. And that’s when we decided that we were going to basically give this away to the people who were the users, let them sell those things in turn into themselves and let them upgrade. We want this really quick installation and we want it to be this instant aha moment, wow this is a really great piece of software and installed very quickly, it solved the problem. Wow, I can see how this could be very useful. Give it away for free. And then they go, they tell their boss we’re gonna go to some more machines well that’s gonna cost two thousand bucks so they make a phone call, the credit card, and up and up and up and so they do that in multiple parts organization and eventually the enterprise license. So, free downloads, free installation and then have been installed themselves. And so, it’s the Freemium model.

Pramod: Yes yes, that’s a great model, yup.

Rob: We were really the people I think, that started the Freemium model for enterprise software. And I think it made all the difference in the world in the way that our company stuck to plan and, and the way that it worked and we still do a lot of that now. But back then, that really was the thing that kick-started us, there’s a combination of software that nobody hadn’t seen before and a very innovative sales model when you put those two together, that’s really, really what made Splunk go. Those two together were the key.

Pramod: Yeah. So, when you give it out there, when the people experience you know that is when they feel they need more and then they upgrade and upgrade.

Rob: Yeah. That’s right and so, but the key is to put a really nice user interface on it because if there are any barriers, they’re going to say, oh! This is complicated and are gonna throw it away.  So, we had to get this what we call 5-minute ahaa. Which means I download it because it looks interesting. (yes), I install it, I can push a couple of buttons on the screen and next thing I know I’m like wow, that may be useful. If you can’t-do that in a five minute, you’re gonna lose them. And I think you asked about what were some of the difficult things. The most difficult thing was getting a consumer-grade user interface on complicated enterprise software designed for operations people.

Pramod: Absolutely! So, now well you had the idea, you had funding, you know now tell us about the growth of the company. From a start-up to empire, like a multinational company today.

Rob: Yeah, I mean it was pretty much constant growth and constant moving from office to office because we were running out of space. Once we started generating revenue, we kept plowing that back into the business and growing so, we grew from two of us to three of us to maybe nine of us over at seven rising funds and gosh back when I left maybe in 2012, I think we probably had five-six hundred people at that point. something like that.

What was interesting was when we actually put together our financial projections when we started the company, it actually kind of tracked pretty well. This was the first time I’ve ever seen something like that happened, there’s this hockey stick thing. We actually followed the hockey stick really closely. Most companies, the hockey stick is optimistic. But it went from a very organic thing, where everybody’s doing everything like this weekend all we’re going to do is just spend on the product testing and file a bunch of bugs. Then we ended up with dedicated teams that started doing things. Then we ended up with as we were making sales. we went from the telemarketing only to we started hiring a couple of people to travel around in different areas to actually try to sell higher-end licenses.

And then we started doing a lot of metrics-driven work? how long does it take a salesperson to get up to speed? What are the metrics on that the downloads versus the installs versus the upgrades versus the all these sorts of things?

So, what we started doing is instead of being very organic and sort of fly-by-night, we started putting more metrics in place and more KPMs in place. on how we manage the company. So, once we started giving these different metrics, you can identify where the holes were. When you’re blind and you can’t see how the business is doing. How do we get more data from those areas where there are holes so that we can have full visibility all the way from product development to all the way to the sales all under the upgrade process. And so, what that did was give us an interesting outlook on how our sales were going to be quarter-to-quarter over time.

And so, we actually developed a pretty predictable sales process, where we could do a pretty good job predicting how much we’re going to get the next quarter, where they going to come from? All that kind of stuff and they’ve done a tremendous job doing that quarter to quarter even as a public company. and I think a lot of that came from the way that we decided to actually use KPMs and metrics all the way through the company.

And so that’s the advice that I would give to everyone is to get literal, factual information on your processes. So, you can use that in an intelligent way. The other thing which I forgot to say is hiring the right executives. We brought Carl I personally think he made a huge difference to the company. Very senior executive and all came together pretty good.

Pramod: Today when you see Splunk, well are you surprised with the success or you had kind of had the idea. It would really go so big.

Rob: Well we certainly were hoping it was gonna get big, we planned for it to get big. I am surprised it did as well as it did. But on the other hand, as we started getting traction because of the people we had, we knew what the right things to do, we knew what the right thing to do was and we knew that our market was really big and we knew that, that type of data, machine-generated data was growing faster than any kind of data, way faster than human-generated data. So, we knew that the overall market for this product was going to get bigger and bigger and bigger and if we could keep our performance up, (yes), and we can keep our user-interface up, and we continue to make good hires that we would be able to still keep up with lot of the demand and we’d be able to create more demand and so I think that was planned right. That part wasn’t a surprise we did that on purpose.

Pramod: It’s awesome. You already answered I think what really went well for Splunk. (yeah), What has been the most satisfying moment for you in the business of the Splunk?

Rob: Most satisfying is watching, generating wealth for all these people for their families, looking at this culture that we built and watching that culture continue. Seeing people that are so happy with their jobs. Watching what happens when you build a company that actually spins off other small start-up companies. I think one of the definitions of a successful company is when people leave that company, start up other companies based on the ideas that they’ve come up with, while they were working with the company like Splunk.

So, I think Splunk is one of these companies where you will see a lot of startups come out of it over time and some are going to be very very successful in what they do, not just because the people are super smart but because they got that spark when they were at Splunk. I think that’s really satisfying to me.

Pramod: Okay…awesome…! Okay are all entrepreneurs originally born as such or are they rise to be successful?

Rob: Oh man…! I think it’s a little bit of both. I think most of it is learned in that when you work in a particular kind of environment, that’s the kind of environment you want to keep working and, (yeah), and if it’s something that made you happy. If your father and your parents were more risk takers and moved around, did different things. I think that you’ll have that tendency later in life. (yes), you’d be able to take those risks. My father wanted me to work at IBM or something like that for our entire career and get a pension. (yes), and I kept telling them look, it’s, there’s this much risk and going to work for a big company as there is a little company. they’re very easy to get laid off during the recession, no matter where you work.

Pramod: Where you work…Yes.

Rob: Yup. And I would you know before Splunk I never worked anywhere more than two or three years. That time I was chasing it sort of the next thing and trying to find the technology that I’m interested in. Oh! that’s cool, that’s quite cool and I wanted to learn it more. if you have a real thirst for knowledge, then especially the software industry where you have the opportunity to move from company to company. You can do that easily here in Silicon Valley. if you are pretty good at what you do everybody wants you.

Pramod: Yes, thirst for change and they want to do something different. Ready to take the risk. So, last question Rob. Do you ever have to deal with the contention with your family?

Rob: Oh gosh… yeah… So in previous companies before Splunk, usually the way we did it is that you know the company would be having problems or something and we’d leave. and get some severance pay or something like that. And then with that money we decided that we were going to start another company and then you know we’d get a little bit of VC money, we’d do it and that wouldn’t turn out so well and then we’re like I’m not putting any money back in the bank at this point in time so, I’m not like building up a retirement. I have a house maybe that I’ve owned because whatever they were cheaper back then when we bought.

But the family is like when you are gonna like to start saving money instead of spending it. When are you gonna have a job that continues to go for a period of time that you can save money so that when we get older, we can retire, all these sorts of things? We get a nice thing we can travel.

Well, when we started Splunk, it was the longest period of time between starting a company and getting funded. So, we went about two years and so for Eric and me, Michael was working for part of that time somewhere else. And for Eric and I, we went two years with no money kind of to that. So, going to 401k money, eating rice cakes, my wife yelling at me all the time when you gonna get a job and I keep telling her okay I have a job. My job is making this company do something. She would say like in past it didn’t. Like I don’t believe you like you really need to get a job and it was getting very very stressful literally. I had no money I was, and my credit cards were maxed out. Eric borrowed out of his 401k it was really, really hard.

Eventually, we got funded, we got paid everything was good, but it put such a, such a stress on my relationship that you know it ended up in a divorce. And I think a lot of it, a lot of it I mean there’s a lot of other issues but, a lot of it was probably because of these stressful financial issues and I really, I really wish that I had more support from my ex-wife at that time. That I didn’t get that I wish, I wish I did. People are different you know, Eric and his wife they get along, she’s very supportive. It takes a special kind of family to be able to work with someone like me or Eric or Michael.

Pramod: Or any entrepreneur

Rob: Yeah, you have to believe in them, you have to believe in them. (yes), You know and it’s tough.

Pramod: And the families got to believe you know, parents…

Rob: Family has to believe. Plus, the number of hours you put in, you’re not there. Doing all these things you’re yelling all the time because something’s wrong. It’s very stressful. And the last thing you need is family stress add it on. (yes). To the start-up stress. (yes). because it’s almost impossible to manage the two at the same time. Yeah, so it was it was tough. Splunk was really tough for me.

Pramod: Yes. And when you look at it should be it such a satisfying feeling now. (yeah), when it’s all goes well. Awesome. Thank you rob it was wonderful speaking with you. So, you took us through a journey of Splunk and on your life. It’s, it was awesome talking to you.

Rob: Thank you very much. I appreciate it.

Pramod: Thank you. Wow, such a surreal experience to speak to the founder of Splunk. Rob, he’s a wonderful speaker as well. A lot of takeaways for an upcoming entrepreneur or a business. I never thought I’ll be sharing the room with Splunk founder when I started this podcast. Absolutely loving it. Next episode I will be speaking to the inventor of structured query language Mr. Don Chamberlin. His contributions are industry standards today. Absolutely curious to know how he designed this beautiful and easy to use the language called Sequel. We all use it every day. He’s also an editor of XQuery and XPath as well. He has made immense contributions to computer science looking forward to speaking to him. Until then. Zai Jian

Leave a Reply

Be the First to Comment!

Notify of