Buddy Brewer (LogNormal)
Joshua Bixby: Hello everyone its Joshua here. Welcome back to the next installment in the Web Performance Today podcast series. I am excited to bring you another chat from my time in London at the Velocity EU Conference. This time I got a chance to sit down with Buddy Brewer. Buddy is the Co-Founder and CEO of LogNormal. I got a chance to sit down with Buddy the day after they were acquired by SOASTA. My good friend over there Cliff Crocker who will be running that division decided that RUM and Synthetic should live together which is a world I have always espoused.
Buddy and I got a chance to talk about Real User Monitoring which is really a hot topic in the industry, I have written a lot about it recently on webperformancetoday.com. We got to talk about how to teach site owners to learn to love RUM and really interesting to talk to someone who took a company from bootstrapping to acquisition in a year. So we got a chance to talk about that as well, I hope you enjoy.
Here at Velocity Europe with Buddy Brewer who has something to tell us, what’s the big news, now it’s going to be late because these pods, this is like the world of podcasting when you have exciting news and then the podcast goes up like a week later, but let’s pretend that the news that aired what a day ago is still modern.
Buddy Brewer: Yeah right at the beginning of Velocity.
Joshua Bixby: Give it to us man.
Buddy Brewer: 8 O’clock A.M. London time and right at midnight back on the West Coast, LogNormal was acquired by SOASTA, so we are really excited about it.
Joshua Bixby: Good work congratulations.
Buddy Brewer: Yeah thanks a lot Josh.
Joshua Bixby: First exit.
Buddy Brewer: It is our first – it’s both my co-founder Phil Tellis and mine it was our first startup that we actually started ourselves, so I have been part of startup before but the first one that we actually did, you know, from inception to exit, so we are super excited about it.
Joshua Bixby: And are you infected with the bug now?
Buddy Brewer: We are.
Joshua Bixby: Yeah.
Buddy Brewer: I got to say we are.
Joshua Bixby: So we got to tell Cliff and the SOASTA guys that what it’s like a two year horizon until you are going to start a new one.
Buddy Brewer: But probably a little bit longer than that --.
Joshua Bixby: See you are still talking the party line.
Buddy Brewer: The beauty of it though is that, you know, SOASTA is not a big company right, so we are going from a tiny company to small to medium sized company that’s had a lot of early success and I think has a really great trajectory ahead of them and I mean we are super stoked that we got to be part of that so.
Joshua Bixby: Yeah I think they are doing some cool stuff. Tell me – give me the birth story of LogNormal?
Buddy Brewer: I think it’s really apropos that we did the announcement here at Velocity because we are a company that really never would have happened if it hadn’t been for the web performance community conferences like Velocity as well as local meet ups. Philip and I actually met in the San Francisco Bay Area, I had just recently moved out there and Philip was living out there and had been for a few years and --.
Joshua Bixby: When is this? What circa?
Buddy Brewer: This was a little bit over a year ago.
Joshua Bixby: Okay.
Buddy Brewer: So LogNormal was incorporated on, if I have this right, it was July 29th of last year.
Joshua Bixby: I love it.
Buddy Brewer: So we are just a little bit over a year old. A few months prior to that, we were going to some of the same events the SF WebPerf Meetup like I said over that they both had in the City of San Francisco as well as occasionally down in the peninsula. And I was looking for something new to do; I just left a company that I have been at for about seven years and I was trying to figure out, you know, did I want to do, I knew I wanted to stay in web performance, but I didn’t know if I wanted to do consulting or if I wanted to go work for another company and I had this opportunity to start something. So after spending all these years in synthetic monitoring, I wanted to spend a little bit of time looking into Real User Measurement, I got really excited about that and that’s how I learned about Boomerang, which was a free library that Phillip had written for people to use to do their own Real User Measurement.
Joshua Bixby: Yeah.
Buddy Brewer: There is just one problem there is no backend for it. So I sat down I said well you know see this is a great opportunity to sit down and write a backend and you know maybe build a nice little sort of business around providing that. So I hooked up with Phillip went and saw him out and sat down and you know figured maybe he’d tweet something about it or something right and you know he says you know what I am thinking about doing the same thing. So once we discovered that we were both on the same track and we were in same sort of geographical area, we sat down and started plan this out and that’s really sort of the beginnings of how LogNormal came about. And we built the product and we got our early thing out there and started to get some interest and thanks to people that we both knew in the web performance industry, we were able to get ourselves into some conversations to people to get some early users that turned into some early customers.
But we didn’t really start to get a lot of exposure until Velocity Santa Clara, that was the opportunity that we had to actually get into a room with hundreds of people and sort of get our perspective on Real User Measurement out there and thanks to the exposure that we got at that conference we really started to kind of ramp our growth to bring on more customers and started to attract the interest of another business that was looking to kind of getting to the Real User Measurement business. And so not only did the web performance community and conferences like Velocity gave us an opportunity to sort of meet and start this company but it also enabled us to grow it and then ultimately was the basis for the introductions that we made that allowed us to exit.
Joshua Bixby: Where is the name from?
Buddy Brewer: Which is the beginning of just something of -.
Joshua Bixby: Which is the beginning yes, the entry.
Buddy Brewer: Yeah right, right.
Joshua Bixby: What’s the story around the name?
Buddy Brewer: Yeah so LogNormal, it’s interesting because Phillip and I knew of each other but we didn’t really know each other super well, so at the same time that we were starting the company, we were kind of getting to know each other’s personalities and how each other ticks. So we went back and forth on the name of our company over and over and over again while we were writing the initial code and you know trying to figure out which law firm to use and all that kind of stuff that you do when you start a business. But the interesting thing about the exercise of naming the business was, it was the first non-trivial kind of not that it was a difficult conversation.
Joshua Bixby: It’s like getting a tattoo.
Buddy Brewer: Yeah I mean I was --.
Joshua Bixby: It’s like getting a tattoo.
Buddy Brewer: Exactly right, so that was a great sort of team building exercise that we went through and we knew that we wanted to start out with Real User Measurement, but you know being that it was a small company and we were just getting started on this we weren’t exactly sure where we were going to go other than that we definitely wanted to stay in web performance. So we wanted to pick a name of the business you know that said something about web performance but wasn’t necessarily or really specific, so we set that on LogNormal because the – when you plot all of the load times in a histogram for any dataset it tends to follow a LogNormal distribution.
Joshua Bixby: Yeah.
Buddy Brewer: So hence that’s the name of our company.
Joshua Bixby: Yeah that’s cool it’s good, it seems very appropriate, it’s like one of those things that feels like it should be named that you know like you hear the name you know that that’s well placed, it’s – I always I liked the name of the business.
Buddy Brewer: And the other thing for us it was kind of like a little bit of kind of like a captured a little bit of kind of this sort of radical upstart notion that we wanted to capture in the way that we were approaching business because we knew that we were small. We were going to be a bootstrap startup; we weren’t starting out seeking capital. So in order to differentiate it was going to be really hard for us to sort of go head to head with a lot of these big incumbent players in the space. So we said you know what we will do is we won’t try to pretend to be something that we are not, we will just try to out geek everybody.
Joshua Bixby: Yeah.
Buddy Brewer: So we will pick the nerdiest name that we can come up with the name of our business and then we will sort of wear that as a badge of pride, so that’s what we did.
Joshua Bixby: Tell me about RUM and you know you guys have been leading this for a year now. Tell me about where it’s been and where it’s going, give me a bit of the history, the Boomerang history and then follow that through to where we are here today and to the future, for people who don’t know about where RUM comes from, give us that story and then pitch us through.
Buddy Brewer: Right, so Real User Measurement first of all, let me preface all of this by saying that I think that there is a place in everyone’s monitoring toolkit for both Synthetic and Real User Measurement. So it’s not an either or proposition, they are both actually quite complementary, but if you look at the history of web performance testing and I have been involved in it in one way or another pretty much since 2001 and when I got started in it really synthetic monitoring was the way to go and one of the reasons why synthetic monitoring worked really well was because there was really only one web browser that mattered and that was IE6.
So you know and that worked really well for a really long time, but then you know some very important things happened right like you know sort of the resurgence of Netscape and then Chrome and Safari and these other browsers on the desktop that entered the field and then the rise of the Mobile Ecosystem. They create all these problems that it was really difficult for Synthetic to deal with, one being browser diversity. It’s just so difficult and also the pace of the revisions of browsers, right. I mean IE6 was IE6 for so long before IE7 showed up. But then you know once that happened and now you have got Chrome like it seems like they released a new version of browser like every week or something.
Joshua Bixby: They are like forty now I don’t know.
Buddy Brewer: Yeah who knows, it’s the things like you know -.
Joshua Bixby: Maybe four.
Buddy Brewer: Maybe we will talk about navigation timing later but it’s easy to remember you know when navigation timing was introduced in Firefox and IE but I couldn’t even tell you, I can’t remember when it was introduced in Chrome because they ran so fast. But you know for those types of things really started to strain Synthetic and especially with mobile where you have got a situation where you know for one thing it’s so dependent on that single strength and the state that the device is in and all of these things that have to do with human activities where they roam around you on the planet and what state their telephones is in and all that kind of stuff that’s hard to capture in the datacenter and I remember in some of their early days when the company I was at was trying to setup mobile synthetic monitoring, they actually had problems getting signals from where they had their phones racked up like they put them in datacenters and things.
Joshua Bixby: Sure, yeah of course.
Buddy Brewer: So -.
Joshua Bixby: That’s a lot of radio frequency disruption when you are sitting in the basement of a datacenter somewhere, right.
Buddy Brewer: Yeah totally, exactly right and one of the things that I was noticing is that a lot of the incumbence were, they had done such a great job building this ecosystem around synthetic monitoring and then they started to add Real User Measurement, but I think where one of the places where Phillip and I wanted to differentiate is there was still this perspective where you lead with synthetic and you did most of your monitoring around synthetic and then where there are gaps like for example these browsers that should you know have covered you over locations which is another huge one, then you would fill in those gaps with real user and where we came in as we said you know we think that that’s backwards.
So we would rather sort of encouraged people to lead with Real User Measurement because there are all these reasons why that’s great, that we can certainly go into and then where there are gaps like for example competitive benchmarking or object level data currently is something that you can’t really get out of Real User Measurement, then go ahead and fill those gaps in with synthetic but our approach was to sort of turn it around the other way and lead with real user. That’s something that people are starting to get now. At Velocity in Santa Clara, we heard so much about RUM; it was great we loved it right.
Joshua Bixby: Yeah and that was good, great timing.
Buddy Brewer: Yeah and this year it seems like RUM is a big thing you know here in London as well as Mobile, which is something that we are also you know very passionate and excited about going through in the future. But the message that we have been trying to get out there around Real User Measurement though especially here at this conference I think is that people are just starting to understand how to use Real User Measurement effectively. So the analogy that I always come back to is, it’s kind of like the early days of television in the 1940’s right, people treated TV initially like it was radio with pictures. So you got this guy that’s you know just reading off of sheets of paper for the nightly news and just reading into the microphone and there is a camera pointing at him and he is not doing anything interesting, he is like smoking a cigarette and reading these sheets of paper right.
Joshua Bixby: Yeah.
Buddy Brewer: And it took a while before people realized that television was actually fundamentally different than radio in some pretty key ways and you could do a lot of interesting stuff you know with those fundamental differences and to bring that back to RUM versus Synthetic I think that what a lot of people are doing now is sort of this first order Real User Measurement where they do the same things that they did with Synthetic like create these lists of my fastest and slowest geographies or my fastest and slowest browsers or you know types of pages, home versus search, you know versus product detail or whatever and that’s important. Those are things that should be done, but that’s to me not the most interesting part about Real User Measurement that the second order things that you can with it or where you sort of take that performance stuff and you correlate it against the behaviors of the real human beings that you collected those measurements from and that’s something that’s fundamentally not possible to do in the world of Synthetic Measurement that’s really powerful when you do it in Real User Measurement because you can discover things from that that allowed you to set goals and make discoveries about how performance relates to your business goals, that was just not possible you know prior to some of the advancements that have happened recently by RUM.
Joshua Bixby: Tell me about it because I really believe that same story, that the power of RUM is that I can connect it to every other analytics, real time or analytics, web analytics tool everything see in Google analytics, everything I see in Omniture can be connected back to time and I can get such a rich dataset. How does this look long term, does Google Analytics and Omniture somehow become the RUM vendor of the future, given that they already have this incredibly rich in user business metric dataset, does it get absorbed there, is it a standalone piece five years from now, where do you, how do you see that evolve for the typical business?
Buddy Brewer: I think it’s going to be really exciting to see what direction that goes in and it’s, I think there is a lot of different things that could happen as far as you know what those businesses choose to do and pull in. But it’s got to start with what I think that we are in the middle of at LogNormal and now are at SOASTA and that’s creating conversations where they didn’t use to happen. Right, so what historically you have had is you have had you know your ops and development people that look at certain datasets from their synthetic monitoring vendors or you know if they are doing Real User Measurement it’s very specific to timing pages and collecting all this technical data. And then you know overall on the business side and the marketing group they are neck deep in their Omniture data and stuff and they like to talk about the metrics that they get out of Omniture and Google Analytics and really have no interest over in the synthetic data and vice versa and they are just speaking two different languages. So by correlating the performance metrics and the business metrics and all of the sudden you are creating this scenario where you know if you relate things like page load times to revenue or to conversions, which is something that you can readily do using Real User Measurement. Then all of the sudden you know the marketing folks get really interested in what those page load times are, the engineering and operations groups I think also got excited about the fact that they can now for the first time understand exactly how you know their inputs to the organization effect the you know the revenues in the top line. So once you create those conversations and then the marketing analytics companies and everybody start to find out that you know that they start to get asked about these types of features.
Joshua Bixby: Yeah.
Buddy Brewer: Then you know I think that for Real User Measurement you will start to see some consolidation and people building things out and you will start to see it happen in both directions. You will start to see more of the performance tools companies getting more involved in that business data and I think you will start to see some of those marketing companies start to pull in more of that stuff that you know it used to not be in the RUM.
Joshua Bixby: Yeah I have been thinking a lot about where that, you know where those worlds collide because they, I think as you said and such a good point the synthetic world has nothing to do with the business metrics right, they are just it’s totally disconnected, there may be some long term correlation between the speed of a synthetic test and performance but there is no direct correlation. What’s so cool about RUM is it’s directly correlated to everything else I am calculating and I think that’s correlation is fascinating. You guys built the business for a year and then sold it why, why not build it, why not and get capital and build it out, what was the choice to join, to merge, get acquired by SOASTA?
Buddy Brewer: Yeah we had essentially I think three different options right you know stay on the track that we were on continue to bootstrap and you know when we needed money to pay our bills and stuff like that while our ramping customers go do consulting.
Joshua Bixby: Yeah.
Buddy Brewer: You know that’s maybe option A, option B is you know go get capital somewhere and go on that track and then option C was to you know to join forces with another company and we were going down that first path and we were starting to see, first of all that people were starting to get that message and as the space heats up of course other people start to come in and the environment gets more competitive and in order for us to continue, to sort of enjoy all of the kind of relevance and mindshare and everything that we are getting, we are going to have to start growing faster and to start building things faster so we are going to need to have some kind of leverage in that right.
Joshua Bixby: Get some capital.
Buddy Brewer: So as we started to move away from that bootstrap model we were looking at capital and we were looking at you know opportunities to be acquired. Really us and SOASTA just kind of like met each other in the right place at the right time and it was one of those things that you know we weren’t necessarily looking at the exact time that we met specifically to go be acquired by someone or have an exit but it was really about the people and we originally started talking about you know sort of this arms linked things about well this could be in anything we could join up with them or maybe we could do a partnership or wherever and the more that we talked, the more that we discovered that we really had a very shared vision around the real user measurement and analytics, so we were really trying to achieve the same thing and that this could be a really symbiotic relationship if we were on the same team because they already had you know this great team of these world class engineers and the sales organization that’s out you know selling their existing products and doing a great job of that and really winning in the market in the things that SOASTA was currently doing and they were looking to get into this other space where we had a lot of expertise, but we just didn’t have that kind of distribution you know channels and everything so by joining forces that it really did create this environment where I think we can continue to do exactly what we always wanted to achieve in the first place which is to get our message about Real User Measurement and everything out there, but do it at a much bigger scale and a lot faster and at the same time you know about this personnel relationships and everything with a group of people that we just genuinely liked and thought that we would have a lot of fun working together over the next you know several years getting it done.
Joshua Bixby: Was there a moment where you thought this is going to be an acquisition, like was there a – did you go out for a phone call or a conference call and you are like this is real.
Buddy Brewer: Yeah, yeah so me and Cliff Croker who you also interviewed go back a little ways and at one point we were talking and he shared a little bit of his roadmap with us and I remember getting off the call and thinking this is the same roadmap, we either need to be competing with each other or we need to be working directly together right you know this is no longer a partnership conversation, this is we are doing the same thing so do we want to do it independently or do we want to do it jointly and again you know like I said it really comes back to the people because we met the rest of the executive team at SOASTA and met the engineering staff and the more people that we talk to and the more we realize that not only did we share the same goals, but this was a great just sort of team to be a part of and a great place to spend the next few years working on technology and really making Real User Measurement happen.
Joshua Bixby: That’s very cool. Any memorable slides from Velocity so far, anything that sticks out you are going to take home?
Buddy Brewer: I would say probably the biggest thing especially here in the UK, probably more so than even Santa Clara is the focus around mobile because you know like I said I spent so much time in the synthetic world pushing desktop stuff and then with Real User Measurement you know just getting our business started and doing all the things that that come out of that has consumed so much of our time and everything that you know I have to admit that frankly I have been probably sleeping on this whole mobile thing more than I should right, even to the point that prior to about a week ago I really knew very little about this whole kind of middle ground where it’s not just about native, it’s not just about HTML5 or pure web applications but there are this whole hybrid -.
Joshua Bixby: Yeah this while hybrid world.
Buddy Brewer: And you know I think that’s been really fascinating, I think it poses a lot of really interesting problems with respect to how you measure performance and how you make those things, how you make those things fast right and that’s something that we are also looking into from a RUM perspective is how we get people insight not only into things that happen in the web browser or a web view but also what happens in the native code and how do we discover a set of metrics that we can use and maybe create some kind of common language where we can you know relate about performance in the native world as usually as we do in a regular browsing world.
Joshua Bixby: Yeah that’s interesting. Lastly, I want to ask about a problem I face which is I go to customers and push RUM a lot as you know I am a big RUM supporter. I face the challenge where customers are so used to synthetic world where they can get an answer in five minutes or they have perceived that they have a viable answer in five minutes and in the RUM world I need to say be patient, you don’t have enough data yet, there is no statistical viability in the data that you have, you need to wait, teaching people about outliers, teaching people about basically stats that we probably all learned in great ‘11 but most people have long forgotten. When you go into organization and they start experimenting with RUM isn’t it a dangerous weapon if used improperly? Do you find the fact that people have to become stats animals, you know animals that think statistically, that that’s a problem for senior executives or other people actually looking at the tools.
Buddy Brewer: Well and there is lies, damn lies in statistics right I mean you can make the numbers say just about anything that you want to make him say. At the end of the day one of the things that we built into LogNormal right from the beginning was putting the margin of error around as many metrics as we possibly could and really trying to give people a sense of the uncertainty around the numbers and be upfront about it. We try to do it in an easy way and we try to, one of the things that we think about a lot and are going to think about a lot more going forward is how we take some of those complex concepts and put them into visualizations and things where people can sort of just inherently understand things about the data without having to, have a degree in statics right I think when you start to confront people with big lists of numbers and all these Greek letters and symbols and stuff that they haven’t seen before all that kind of stuff and yeah -.
Joshua Bixby: Yeah here is the standard deviation and it’s like you have lost all bunch of executives at that point right.
Buddy Brewer: Yeah it really starts to scare, yeah, yeah exactly so I think that the biggest challenge with anything that’s data oriented is finding ways to conceptualize the data and to present it in a visual way that communicates some kind of useful message without over simplifying and giving people the wrong impression.
Joshua Bixby: And there is that balance right?
Buddy Brewer: And that’s one of the things that you know I think some of the early tools and one of the traps that people fall into a lot is you know allowing outliers to secure their numbers and they put these you know numbers up in great big huge giants fonts and they are just not very realistic for all sorts of problems right. So we try to use, it’s one of the reasons why we tend to focus more on percentiles and putting things in histograms.
Joshua Bixby: Yeah.
Buddy Brewer: You know rather than showing arithmetic means and stuff like that and also why we really try to be upfront with people and in fact there are some metrics that we don’t even show unless we have you know a minimum sample size.
Joshua Bixby: See that’s what I want, that’s what I have always pushed forward, don’t even give, don’t even give them the gun you know until, until it’s like you know until you are in a position where you know it’s safe, just don’t even give them the tool because they are going to make bad judgments. The tools can be an interface to help educate us on when judgments can be made and when they can’t and when we all see a number we make judgments right?
Buddy Brewer: Yeah and that was one of the things too with kind of going back to the why SOASTA thing you know that we got really excited when we were working with those guys because we had seen some of the visualizations and the UI work that they had done in their existing products, so it wasn’t just about getting any old group of the bunch people and just having them you know help us scale our business. It was about you know this very specific set of expertise around, I think especially around the visualization side where we could learn things and apply that expertise to Real User Measurement and accomplish some pretty interesting things.
Joshua Bixby: I wish you all the best of luck, we will be in touch. I appreciate you taking the time. Are you going to go out and see a bit of London now?
Buddy Brewer: Yeah I think so I hope Cliff has been waiting on me -.
Joshua Bixby: I know, I know I got to get you out of here because this is Cliff’s first time to Europe so you guys have to get out, the clouds are coming in, enjoy, have some good beer and let’s stay in touch.
Buddy Brewer: Thanks a lot Josh.
Joshua Bixby: Okay take care.
Joshua Bixby: Thank you for taking the time to listen to today’s podcast. Many thanks to Buddy who took time in a very busy Velocity Conference, for him a very busy schedule, he took time to sit down and visit with me. I wish him all the best of luck and I am very excited about what they are doing over there at SOASTA. Check out the rest of the podcast in the series at webperformancetoday.com/podcast and if you have any suggestions for future topics or even better, if you would like to volunteer yourself to put yourself forward, to speak out on performance, on web performance, acceleration any topic related please reach out to me at firstname.lastname@example.org, I have been really enjoying these conversation and look forward to talking to more really interesting people. Have a great day.