Tyler McMullen is CTO at Fastly, a global edge cloud platform. As part of the founding team at Fastly, he built the first versions of Fastly’s Instant Purging system, API, and Real-time Analytics.
As such, he was also part of Fastly’s 2019 initial public offering (IPO). Fastly’s solution is another example of the growing importance of the cloud for businesses.
“Cloud computing’s importance has grown significantly in recent years,” as a recent GlobalData research report noted. “It has enabled the use of shared IT infrastructure and services to create a flexible, scalable, and on-demand IT environment.”
In the latest Q&A in our weekly CTO Talk series, the Fastly CTO explains how he ended up in his role, what advice he’d offer other CTOs and talks about the ethics of technology.
Tell us a bit about yourself – how did you end up in your current role?
It was about 10 years ago now that Artur Bergman, the founder of Fastly, approached me about the beginnings of Fastly. I was sleeping on a mutual friend’s couch, and had spent a number of years working on distributed systems. It was a really interesting area for me. Artur asked me about making a CDN, but caveated that it’s more like an edge computing network. At the time, that was like catnip to me. A globally distributed system is far more fascinating than other stuff out there that I was looking at the time.
It is extremely challenging and requires you to rethink norms. So, we started there, and as is the case with startups that have a useful product from the beginning, we launched with what was essentially a smarter CDN. We gave people the ability to write some level of code at the edge, and our goal all along was to make a much more full-featured development environment and platform out of it. Not just for content delivery: we wanted to move the thing that’s generating the content to the edge.
Where did your interest in tech come from?
I got involved in tech roughly the same way a lot of kids in the 80s and 90s did, which is through interacting with my Nintendo. Video games were one outlet that got me into the tech world back then, and what really amplified that was my first Game Genie. In playing around with that, I had the realization that what the Game Genie was actually doing was modifying memory locations within the game. So, it was a realization that it was just memory and addresses within a linear space, and it clicked for me. These seemingly magical experiences I had with video games weren’t as complicated and magical as they appeared – you could actually break things down and figure out the way they work. That experience was really fundamental for me and made it clear that nothing is magic and if you think about it hard enough you can figure out most any technical problem. Even the most complex applications are just carefully stacked towers made of fundamental primitives.
Which emerging technology do you think holds the most promise once it matures (and why)?
Obviously I have to say edge computing. It turns the entire network into your client, server, datacenter and it unifies all the different models of network interaction. It also solves real problems faced by developers in the modern world where much of development is making many different services work together effectively and efficiently.
How do you separate hype from genuine innovation?
You need to ask whether it solves a real problem, and does it do so in a way that is flexible and broad enough to apply beyond the first concrete example? You also need to ask if it is efficient? Lots of problems can be solved by applying inordinate amounts of resources to it. It’s only interesting when the solution is more efficient than living with the problem.
What one piece of advice would you offer to other CTOs?
You must learn to accept that you can’t solve every problem. You need to pick what you care most about and run it down. Trying to do everything leads to burnout and ineffectiveness. This is harder than it sounds as you’ll be pulled in dozens of perfectly reasonable directions. I would suggest that you also have an opinion on a situation but be willing to change it. Flexibility is key for fast moving situations.
You must also make decisions. Sometimes they’ll be the wrong ones. But a wrong decision gets you to answer infinitely faster than no decision. It’s a trope at this point, but the statement that no decision is still a decision is accurate. Don’t let your choices default.
What’s the most surprising thing about your job?
Well, it’s not surprising to me, but some people may be surprised to know that my role and the office of the CTO at Fastly looks a little different than what many think about CTO roles. We spend time on a couple different kinds of projects: “That’s Weird” projects and ambitious projects with a high probability of failure (but a high upside if they work). “That’s Weird” projects are when we see something just slightly odd. We try to keep an eye out for unexplained phenomena and try to run them to ground.
What’s the biggest technological challenge facing humanity?
The biggest challenge in my mind is actually in the ethics of what we build and how we choose to use it. We frequently see technology used to invade privacy, violate the rights of groups of people, and otherwise harm those with little or no recourse. Technology companies have a tremendous amount of power, and the engineers at them wield much more power than they realize. We as an industry and society are at an important juncture wherein, we’ll decide the kind of technological future we want.
What’s the strangest thing you’ve ever done for fun?
I used to restore all broken down dirt bikes for fun. It is a surprising amount of fun to take a machine that has been neglected and broken down for years and turn it back into something that people can enjoy. Also, using Coca-Cola to release a seized engine really does work.
What’s the most important thing happening in your field at the moment?
Several things jump out at me here. Early models of edge computing were extremely varied regarding the execution model. The trend now however is that edge computing is almost necessarily serverless in nature. This points to the realization that edge computing is again, not just about a specific location in the network, but rather about being able to place computations where they are most needed and most efficient. The serverless model makes that possible. As soon as you land on the serverless model however, another problem makes itself apparent. That problem is language choice.
To my mind, the next nascent trend is around state. Edge computing by itself is great and powerful, but without access to low-latency, stable, and standards-based storage and state engines at the edge, the use cases will necessarily be limited.
In another life you’d be?
I was big into skateboarding when I was younger and had dreams of going pro. It was only when I failed at that that I decided to go into technology. So, in another life I’d be kick-flipping down massive sets of stairs instead of writing code.