Josh Fraser (Torbit)

Podcast Transcription

Joshua Bixby: Hello everyone. Joshua Bixby, President of Strangeloop Networks here with another edition of the Web Performance Today podcast. This week I’m talking with Josh Fraser, CEO of Torbit and given that they also do some FEO, I supposed my mortal arch enemy. We talked about discovering the performance path, how performance has changed from a technical metric to a business metric, which is a topic that as you know is very interesting to me, and talked a little bit about some advice for young entrepreneurs, and finished off with Josh’s predictions for 2013. Josh, I consider him a friend and a colleague in this industry and I hope you enjoy the chat.

Joshua Bixby: I’m here with one of the dynamic CEOs in our business and my sworn enemy, right Josh? Josh Fraser, CEO of Torbit. Josh as I was doing this podcast I was talking to one of our joint customers actually. Friends over at Wine.com who are big believers in what you guys are doing from a RUM perspective and he’s like “dude you can’t talk to Josh, you guys do the same thing” and I’m like come on man, it’s a big big world so Josh, welcome to the podcast. I consider you a friend and it’s nice to have you here. How are things going?

Josh Fraser: Things are going well and I totally share the same sentiment as far as we both have the same goal of making the internet faster and so obviously support what you guys are doing and glad you reached out and we’re getting a chance to do this podcast.

Joshua Bixby: Yeah, it's nice to chat, last time was probably Velocity last year I guess would have been the last time we saw each other. Is that right?

Josh Fraser: It’s been it’s been a while yeah. I know you were at one of the meet ups.

Joshua Bixby: Yeah, the San Francisco one, that’s right.

Josh Fraser: Out here in the valley. I remember going out for drinks.

Joshua Bixby: That was fun.

Josh Fraser: Yeah, so that was a great talk you gave there, certainly enjoyed that.

Joshua Bixby: So I want to hear about I want to hear about your journey. So you’re obviously the CEO of Torbit, which does front-end optimization as well as RUM and I want to talk about Torbit but I want to hear your pedigree before Torbit. What got you into the web performance space? Why web performance?

Josh Fraser: Yeah well so this is my second company and my last company I was working on was called EventVue which was not really into performance at all, we were doing social networking around events but one of the things that people would complain about with the application was that it was really slow and people would sign in and say you know “I love this service but it’s too slow. I can’t use it” and so that was sort of the first spark that sent me on this path of trying to learn all these tricks for performance optimization. Stumbled upon Steve Souders work, started learning all of these tricks on the front-end that I had been completely oblivious to myself. At that point, I started to really enjoy it and it became a game to me of how fast can I make it go but at the same time, it got to be really tedious because we’d do site redesigns and keep pushing out changes to our site and you know every time we had to go back through and manually optimize the images and combine the CSS files and the JavaScript files and all these optimizations which were really tedious so it sort of struck me at this time like hey this sort of stuff, this is what computers are made to do. They’re better at this than humans. I was busy with EventVue and didn’t really do anything with that. It wasn’t until sort of after EventVue shut that down that looking around for my next gig so the idea came back around performance and met up with another technical guy and he kind of brought the idea up and said why don’t we automate that and that was sort of the first spark that got us interested in the performance space.

Joshua Bixby: I love it man. You know you’re preaching to the converted when it comes to the story of trying to make things faster. Tell me about you know the founding of Torbit. I know you guys moved from Boulder into California. Tell me about just the Boulder to California move. Why did you guys decide to locate in California, obviously talent but was there anything else that brought you there?

Josh Fraser: Well I think we wanted to make a pilgrimage to Mecca. You know, being in the tech world, it made a lot of sense to us as far as a lot of engineers out here and just personally wanted to spend some time out here to build my network with people here. I’ve certainly found out here that the network goes a lot deeper and the expertise goes a lot deeper here as well I’ve certainly enjoyed being here in California. Not sure yet whether I’ll stay permanently or not but you know it was both for personal reasons as well as Torbit and thankfully when we made the move, it was myself and Jon at the time so it was a somewhat easy transition to move the company.

Joshua Bixby: I’m interested in what you’ve seen change over the last few years. You’ve been in the performance market over some very critical years as we’ve seen evolutions in browsers, and in proliferation of mobile. What have you seen change in three years – probably the most important three years in the history of the industry. What have you seen change?

Josh Fraser: Yeah I think one of the biggest things I’ve seen change is performance changing from just being a technical metric to a business metric. I think we’ve really seen that happen, particularly in the last year where business after business is starting to pay attention to the economics of it and performance is something that not only the engineers care about, but you know CTOs, CEOs, VPs are starting to really be aware of performance and really taking it seriously. And people are evaluating it not just on sort of how fast the site is but how much money is at stake and making a business decision on how much to invest in your CDN and your performance engineers and of other technologies like Strangeloop or Torbit site optimizers.

Joshua Bixby: I’m interested in your, in the move into RUM and the information that you’ve been able to gather there. Has this changed your perspective on performance? I mean you guys must capture more RUM data or as much RUM data as anyone in the world at this point in terms of breadth – the number of customers. I mean you saw I did a blog post recently where you guys came up I mean right in the top RUM collectors. Has this changed your perspective on performance?

Josh Fraser: It has. I think what we realized is that accurate measurement is step one and you just can’t, you know we don’t even want to talk to someone about optimizing their site until they have accurate measurement in place. So you know the way we’re looking at it right now is let’s get everyone in a row using our analytics and really understanding how fast their site is. There’s really no point in talking about anything else until we have that measurement is in place. It’s really hard to lose weight when you don’t own a scale.

Joshua Bixby: Yep

Josh Fraser: And so you know, our view is let’s get as many people getting their eyes on this data and that alone if we do nothing else that alone will help make the web faster. We’ve seen that already.

Joshua Bixby: It’s interesting because as you know we’ve also been big proponents of the numbers game. Taking a bit of a different approach in the sense that we’ve chosen to really embrace tools like yours and others on the RUM side as opposed to developing our own. I just had a really interesting chat with Eric Goldsmith where the conversation was about how do you make RUM actionable or part of what he’s working around is how to make RUM actionable where the signal and noise challenge is so significant if you see a bad result it could be because the guys cable modem is down, it could be because you know the CDN didn’t work, it could be because so many issues and he’s trying to find a way to collect enough information to signal ops. Have RUM be a part of signaling ops because traditionally in the synthetic world the test gets bad, two or three tests in a row, ops get an alarm. There’s probably something wrong, you’ve controlled enough of the variables that there’s probably something wrong. Have you seen how are you seeing ops interacting with RUM? That’s an area that has been intriguing me and then that conversation with Eric certainly increased my interest in the topic.

Josh Fraser: So one thing we’ve seen is we just added multi-user capabilities to Torbit and that’s being used a lot and it’s across different teams. So ops are using it, we have front-end teams using it, we have business decision makers using it so I think one of the interesting things is seeing different teams across the company really using it. I think as far as ops go, I think you’re right. The challenge of alerting is really hard because you’ve got so much data and you know what is considered an anomaly changes from site to site. It changes by time of day, by day of the week, and even seasonal changes things, too. Just looking for you know this changed 20% from an hour ago isn’t necessarily good enough. You have to be a little bit more intelligent than that.

Joshua Bixby: Yep.

Josh Fraser: And so our first attempt it we’re working on some of the key features now, and sort of the first version of that is just you know, letting people set their own thresholds on that, right. Completely hands-off approach of “you tell us if x variable changes by x percent, send so and so an email.”

Joshua Bixby: Yah, makes sense.

Josh Fraser: And I think that’s a really simple first step, but the aspirations are to get a lot better than that. I’ve actually been a lot smarter about it, using a lot of statistics and probability and these sort of things, as well as the massive set of data that we have making intelligent, you know sharing information with them where it’s like hey, your site is slow, but don’t worry, everyone else’s site just slows down too it’s something on the internet backbone, you know here’s a link to a news story about someone who dug up cable in the wrong place or something like that.

Joshua Bixby: Yah you’re hosted at Amazon too or whatever, I shouldn’t say too much cause they’re, you probably saw they’re a recent partner, but wherever you’re hosted you know our other 18 customers on that are also down so you know as you say, don’t worry about it. But that’s been something that’s really interesting, has been interesting to me. Another thing I’d love your perspective on is the future of this. I’ve, in the post that I referenced where we saw your beacon come up quite a lot, I also really discounted Google Analytics as a RUM solution. I mean I did a very informal study and found most of my customers know it’s there but have never used it. When I roll this out three or four years, is RUM a part of a business analytics tool, like Google? Is RUM part of a synthetic tool? Does this all come together, where do you think this ends, or what’s its progression? Because it probably never ends.

Josh Fraser: Yah I think, I mean I think Google Analytics is a great tool, a lot of people use it. And what’s nice is that you can do these correlations back to, you know they already have conversion data in place. So some things are really easy for them to do it on that. I think there will always be space for people to go a lot deeper into this particular thing. And we’ve really tried to keep our focus just on performance. We don’t tell you where your traffic’s coming from, we don’t do any sort of other analytics other than performance. And I think that’s important because it lets us go really deep on it. One of my biggest complaints with Google Analytics is that they use averages for everything.

Joshua Bixby: I know I’m the same way man, I’m the same way.

Josh Fraser: It’s, you know, the case studies from Walmart were for years they thought their loading time was 4 seconds because everybody had told them the loading time was 4 seconds, and it wasn’t until they built their own RUM tool and rolled that out they found out you know the 90th percentile was 20 seconds, and Google Analytics was one of those offending products that tells them hey, your load time is 4 seconds don’t worry about it. You know there’s other problems as well, I mean they use just the browsers that use tiny with no support for older browsers at all, and you can’t see the head scan, you can’t see top percentile, some of these things we think are really basic to getting an actual understanding of what’s going on.

Joshua Bixby: I want to switch gears quickly. I know as a young CEO executive, both of us are in very enviable positions, I know some of the, and I get this question and I’m sure you do quite often, I get the question, how have you started businesses and I know you’re a big part of both participating as a mentor and being mentored. Tell me about starting a business, and the mentors you’ve had, the advice you’ve received. I’m interested in the starting business side of your world, so the not the performance side, but the other side. What advice do you give people about starting businesses, whether it’s in this space or others. I’m intrigued because I think you’re one of the really successful ones that have come up and I envy that, so I’m interested in the advice you’d give to people.

Josh Fraser: Yah well someone gave me some advice a long time ago, and they said “always try and surround yourself with people smarter than yourself.” And I’ve been successful at doing that, it’s a pretty low bar ha ha. You know I’m definitely the dumbest person we have at Torbit. And I think that’s been, that’s something that we continue to strive for, to bring on you know the smartest people we possibly can. And I think that, you know, to be a good CEO you need to be always reaching out and relying on other people’s expertise, particularly when you’re a young CEO and you’re doing a lot of stuff for the first time, like going to a new skill that you’ve never seen before. It’s crucial to have all those other people around, those mentors you can go to and not being afraid to tell them that you don’t know something and ask for help, and that’s something that we try to do and something that’s a core value at Torbit.

Joshua Bixby: Is that easier in the valley for you, than it would be somewhere else? Or harder, I mean there are more of you, there are more of us.

Josh Fraser: Yah there’s certainly more people, the expertise goes a lot further back. Right I mean you can go talk to some of our technical advisors are if you people who were at Netscape in the early days. Lou Montulli is on our advisory board, he invented cookies, for example he invented the blink tag. And he’s on our advisory board right, we get lunch with him every couple of weeks. And so getting that sort of expertise around is really invaluable, and I think it’s one of the reasons to be here in the valley. That said, sometimes being in smaller communities has advantages, too. When we were in Boulder, people were even more willing to, you know, just go grab coffee with you and just get to know you, whereas here the valley is go-go-go, you know, what do you want to meet about, you know send me an outline of what you want to discuss and I’ll put 15 minutes on my calendar. And so I think it’s a little bit of different culture in that regard.

Joshua Bixby: And if you were to do it all again, the next business, and you could have the beauty, the comfort, the ease of Boulder – as an example, which is a beautiful town, my sister-in-law lives there so I get to spend a lot of time up there – or in Mecca, where would you start now, knowing what you know?

Josh Fraser: I think, honestly I think for your first, my advice to someone that’s thinking about starting a company, I think for a first company somewhere like Boulder is a great place to do it. For, you know a second company, for Torbit, the valley is where we need to be. So I think it really depends on the type of company you’re building, and what makes sense, obviously you guys are up in Canada, that’s working well for you, so I don’t think you have to be in Silicon Valley, there’s just pro’s and con’s. You know, rent is really expensive here, gas, everything is really expensive.

Joshua Bixby: Talent?

Josh Fraser: Yeah the only thing that’s cheaper here is wine.

Joshua Bixby: Oh man, and such good wine, such good wine. Oh, you’re so lucky. I want to know about a few things that are going to surprise me next year. What do you have on your horizon? Not Torbit-specific, but what are you seeing in the tech sector, related or even unrelated to us that are going to be surprises in the new year.

Josh Fraser: Yah well I think resource timing it’s going to be interesting to see where that plays out. I think there’s, you know, that’s going to be an interesting thing to see you know how much adoption that gets in the browsers, what that means for synthetic testing, I think sort of being able to get down to resource level is one of the biggest reasons people are hanging onto their Keynote accounts. And so if they get, it has a lot of potential, I think I know as far as what’s actually going to happen there. I think we're going to continue to see destruction on the CDN stage, I think people are starting to realize you know the CDNs are focusing more on their revenues, then they are on making you fast, and are doing a lot of CDNs are doing a lot of things to hurt performance to increase you know their revenue. I talked to a customer the other day, $1.4 million dollar contract with Akamai, and they found out they turned off tzip, and you know the result is they’ve blown through their contract in twice the time and you know obviously they weren’t very happy about that. But I think people are starting to, I think one of the big sort of shifts I’m seeing is sort of a lot more awareness in the market, a lot more people sort of waking up to the things that have been going on, and I think there is a lot of disruption coming in that phase as people realize oh wait, Keynote has its boxes sitting right on top of Akamai’s, maybe you know we can’t trust these numbers as much as we thought. And so I don’t know exactly what is going to happen there, but there’s disruption coming to that space for sure.

Joshua Bixby: Yah that’s interesting I know I’ve thought resource timing is something that the CDNs really don’t want to have out there either because it’s, that could be a scary thing for a lot of my customers who believe they’re at the edge and cached 100% of the time, where the reality is that they’re not the NFL and they’re not Netflix or whatever the case may be, or some big content site and they’re not at the edge, they’re somewhere deep inside a CDN’s network and it takes them as much time or in some cases longer to get from there than from their origin because of all the checks. So I hear you, I think that’s really interesting. Mr. Fraser, I want to get you back to your domain, let us continue to be friends and foes, and I recommend everyone to go out and take a look at what Torbit’s doing, as you know I’ve been pushing our customers in that direction, too. So keep up the good work, it’s a pleasure to chat, and I will see you probably in a few months.

Josh Fraser: Sounds great, and well thanks again so much for having me, Josh.

Joshua Bixby: Of course, take care.

Joshua Bixby: Thanks again to Josh for joining me, and to you for listening. If you want to hear about some of the other really interesting conversations we’ve had on this podcast series, please go to webperformancetoday.com/podcast. I’m also always eager for suggestions for guests, topics, so feel free to reach out, joshua@webperformancetoday.com. Have a great day.

 

Where to find Josh:

Mentioned in this podcast:

Subscribe in iTunes: