That adult was Paul Sagan. Sagan had recently moved with his family from Switzerland, where he enjoyed a prestigious, yearlong position as a senior advisor to the World Economic Forum. He came to Cambridge to find the next step in what had already been a wild ride in the news business. Born and raised on Chicago’s South Side, Sagan’s career in journalism began at the age of thirteen, when he earned $25 from a newspaper for a photograph he’d taken of a local car crash. He came by the profession naturally as the son of a newspaper publisher who’d always been interested in the way technology and new media shaped the news. With a degree from Northwestern University’s Medill School of Journalism, Sagan joined WCBS-TV in New York where he began writing news at age twenty-two. “I was horribly underqualified,” remembered Sagan, a bespectacled, tall businessman who is sharp-witted and affable. A serious, earnest young reporter, Sagan was the only staffer who dressed in a three-piece suit and tie at the office. He quietly but diligently worked his way up to a spot as the youngest news director in the history of the network.
After a decade at CBS, Sagan was recruited by Time Warner to start a news channel in New York called NY1, which he built from the ground up. He also helped to lead the company’s online efforts with the launches of Roadrunner, the first cable modem service, and Pathfinder, one of the first Web sites in the world with advertising. In 1997, in his third and final job at Time Warner as president and editor of Time Inc. New Media, Sagan decided to make a change. By then, he had three children with his wife, Ann Burks Sagan, and he wanted to spend less time at the office and more with his family. When the stint at the World Economic Forum helping to plan the exclusive Davos conference came along, Sagan jumped on it. But after a year away from the hustle of a news-driven office, Sagan was ready for the next step. “I always loved the pressure of live TV news, especially the glory days when we did broadcasts at noon, five, six, eleven p.m.,” he said. “And, at the time, a lot of people were saying the Internet was just as competitive and crazy.” He began talks with some venture capitalists in the Boston area, starting with Battery Ventures, which he had connected with while working at Time Warner. Sagan was introduced to Dagres and Scott Tobin of Battery, who sent Sagan to “check out this crazy business plan” called Akamai.
Sagan did his due diligence. He read their crude business plan and Power Point presentation and met with Leighton, Lewin, and Seelig in Leighton’s office at MIT, instantly taking a liking to the group. “They were super nice and super inquisitive,” Sagan recalled. “I grew up in Hyde Park surrounded by University of Chicago academics”—his second cousin is the famous astronomer Carl Sagan—“so for me it felt like a natural fit.”
Sagan took a clear message back to Dagres and Tobin: “I told them that I was no engineer, but that it looked to me like they had a working system. And if it works in the wilds of the Internet, they have a big idea worth investing in.” Dagres trusted Sagan, but not everyone at Battery was convinced. “There were some doubts within my firm. One of my partners claimed that algorithms were not enough to build a company,” noted Dagres. “But I really had faith in the people. I felt they knew something special and they’d work tirelessly to achieve a mission.” To bolster his firm’s confidence in Akamai, Dagres took an interesting proposition to Sagan: “We’ll invest, if you’ll go watch the money,” he bargained. This because, as Dagres reasoned, “Sagan knew how to run a company, [and] he understood the media space and [was] smart and had high emotional intelligence.” Sagan agreed to a six-month stint helping the theoretical mathematicians bolster their list of contacts, improving their business skills, and, of course, overseeing the money. He needed a title that would grant him the authority to make big decisions, so he took on the role of chief operating officer (COO). At that point, Dagres said Battery had the confidence to invest in Akamai: “You never know for sure, but you take a shot.”
With the belief that the big venture money was soon to come, Leighton signed the lease on a new office. Akamai moved from its tiny office space at One Kendall to the fourth floor of 201 Broadway in the heart of the Cambridge hub for high-tech firms. By November, everyone involved in the first-round financing rush was anxious—a deal was on the table, and it was a good one. Two prominent firms—Battery Ventures and Venrock{§}—were ready to make an investment of $8 million ($4 million each) in Akamai. The deal was as good as done, signed, and delivered to Akamai’s financial officers and attorneys for vetting. But on a Sunday, the night before the funds were to be wired into Akamai’s account, Leighton’s home phone rang. It was one of the general managers of Venrock. “We’re sorry,” he told Leighton. “We’re not going to close the deal.” Leighton was stunned. Venrock seemed like the perfect fit, having been an early investor in both Apple and Intel. The partner went on to explain: “I’ve thought hard about it, and I don’t think you’re going to be successful.”
Leighton could only think of one thing to say in response: “I think you’re wrong.”
In one phone call, Akamai lost half of the $8 million they’d been promised. Lewin immediately feared a domino effect after Venrock’s sudden departure. Having turned down offers from almost all the local venture capital companies, they worried no one would even consider stepping in if Battery backed out, too. Dagres reassured them, though, offering to close the deal if Battery could take charge of finding another investor to fill the hole left by Venrock. Akamai agreed. Their best hope was Polaris Venture Partners, which had recently recruited George Conrades, a veteran from IBM and BBN (Bolt, Baranek Newman) who had long governed the intersection of business and cutting-edge technology. Dagres contacted Polaris with a deal: if the firm would supply the $4 million Akamai needed, the company would offer Conrades a position on its board. A deal was struck, and in November 1998, just before the Thanksgiving holiday, Akamai secured its first round of financing from Battery and Polaris.
Leighton and Lewin retained control of the company with a stake of $8 million. Lewin took the title of chief technology officer, and Leighton became chief scientist. Seelig assumed the role of vice president for Strategy and Corporate Development. Kaplan became vice president for Business Development.
After the deal was signed, everyone went home to celebrate the holiday. As for Venrock’s last-minute decision, the true cause leaked later: the deal fell through not because Venrock lost faith in Akamai, but because a senior partner in the firm’s New York office had butted heads with Lewin. “He thought Danny was arrogant and didn’t like his demeanor,” said Dagres. Venrock’s decision to pass on the deal proved costly: if the firm had taken a 10 percent stake in Akamai for $10 million, it would have been worth $2 billion less than a year later.{34} Dagres said it went down as “one of the dumbest venture moves in history.”
A young analyst at Venrock, who had worked around the clock to set the deal and, in the process, came to admire Lewin, sent him a book the week after the breakup. It was the inspiring story of a business mogul. Under the cover flap, he inscribed a note to Lewin: “You’re a titan!” Lewin seized on the word. Not because it was used to praise him, but because he loved the ring of it. Titan! A word with mythological Greek origins that had also come to describe anything with a massive, indomitable spirit—from one of Saturn’s moons to sports heroes and business magnates. It was pithy and powerful. Lewin kept the book in his office, and from that day forward, the word “titan” became a part of Akamai’s vernacular, and it was reserved for the select few who exceeded even the highest expectations.
Photo Section
Danny Lewin as an infant with his mother, Peggy.
Lewin family
Lewin with his younger brother, Michael.
Lewin family
Lewin as a young teenager in Israel.
Lewin family
Danny and Anne on their wedding day—August 20, 1991—in Jerusalem.
The pair wed in a small ceremony attended by family and a few friends.
Anne Lewin Arundale
Anne and Danny with their two s
ons, baby Itamar (with Anne) and Eitan
(with Danny) on a trip to Israel in the late ’90s.
Anne Lewin Arundale
Lewin working out at Samson’s gym in Jerusalem.
Lewin family
Lewin and Leighton (with his back to the camera)
huddled with co-workers for a meeting at Akamai in 1999.
Akamai
Lewin as a soldier in the Israel Defense Forces;
he served as an officer in the army’s
most elite counterterrorism unit, Sayeret Matkal.
Lewin family
Marco Greenberg (left), and Lewin by the Western Wall
in the city of Jerusalem, where the two first met and became best friends.
Marco Greenberg
Tom Leighton and Lewin in front of one of the white boards
they used to draft algorithms at MIT’s Lab for Computer Science.
Chia Messina
(From left) Leighton, Lewin, Paul Sagan, and Jonathan Seelig
in 1999, shortly after Akamai was founded.
Akamai
Lewin and Paul Sagan at a data center at Boeing Field in Seattle,
using a tool kit to reassemble Akamai servers that crashed
during the company’s first month of business.
Paul Sagan
Akamai Technologies headquarters in Cambridge, Massachusetts.
Akamai
Akamai’s Network Operations Control Center,
from which the company has a bird’s eye view of global Internet traffic.
Akamai
Lewin on one of the many motorcycles he purchased with his newfound
wealth after Akamai’s successful public offering in October 1999.
Laura Malo
An iconic shot of Lewin and Leighton on the steps of MIT’s
great dome in 1999, just after the formation of Akamai.
The Boston Globe
Danny Lewin’s name on the 9/11 Memorial in New York,
at the site where the World Trade towers once stood.
Molly Knight Raskin
Chapter 7
Science or Snake Oil?
“Science is magic that works.”
—KURT VONNEGUT,
Cat’s Cradle
AKAMAI HAD MONEY in the bank—lots of money. Despite the fact that the company still existed only on paper and in a student-built prototype at MIT, it was flush with cash. But to build the actual infrastructure, one formed by thousands of servers around the world, they would have to spend the bulk of it. Akamai had the space to install the servers through their commitment from Exodus. The company still needed to buy the servers, however, and somehow install them in the Exodus centers, some of which were located in remote corners of the globe. At the same time, they needed to sign up some content providers to try the Akamai service once it was up and running. To make all this happen quickly, everyone at Akamai again turned to connections. Lewin, Leighton, Seelig, Kaplan, and Greenberg scrambled to set up more meetings with friends, even friends of friends. They combed the MIT alumni database. They asked their investors to make calls and open doors on their behalf. They hopped on planes to meet with almost any interested party who offered them a few minutes of time. The hustle paid off.
Through a contact at Exodus, they found a San Diego–based company, True Solutions, to make four hundred servers. At the time, servers were not cheap, costing upwards of $10,000 apiece. But True Solutions was a little-known company without much of a track record, so they promised to quickly construct the servers to Akamai’s exact specifications at a bargain price. Each of the servers would run on an operating system called Linux and would be programmed with software driven by Akamai’s proprietary algorithms. Within a few weeks, True Solutions had already churned out approximately three hundred servers, and began shipping them out to eleven Exodus data centers in nine locations. Of those nine, only one of them was outside the U.S. But to Akamai, one server in London was enough to call their network international. With the servers in hand they needed to get them securely into the data centers. Exodus employees installed as many as they could, but left the rest to Akamai.
On a bitter December day, Lewin and Seelig drove out to a data center in Waltham, twenty minutes outside of Cambridge, to install one of Akamai’s first racks of servers. “We literally threw these boxes of servers into the back of my car,” said Seelig. It was after 4:00 p.m. when Lewin and Seelig pulled into the parking lot of the massive center located on Winter Street at the heart of the Route 128 tech corridor. Almost as soon as they arrived and began unloading, Lewin realized he’d forgotten their tool kit. They considered turning back, but they anticipated a few hours of work, and it was getting darker and colder. At that point, they decided they had only one option: the spare tire repair kit in Seelig’s Mazda. They grabbed it, hauled their boxes into the building, and hoped for the best. It took four hours for the two of them to manually screw the servers into the racks, and by the end their fingers were throbbing. From the parking lot, they called Akamai to confirm a signal. The servers had been successfully connected, so Seelig and Lewin triumphantly headed home.
On the way, they spotted an Italian restaurant by the side of the road and decided to celebrate with a meal and a bottle of wine. “We were really excited,” recalled Seelig. “Neither of us had heard of the place, but it looked good and we were hungry.” It turns out they had stumbled on La Campania, a top-rated, pricey establishment popular with the venture capital partners in the surrounding offices. Clad in grease-stained jeans and covered in bits of cardboard and Styrofoam from breaking down dozens of boxes, Lewin and Seelig were hardly dressed for the occasion. “Danny and I were filthy,” noted Seelig. “So we were a little uncomfortable when we sat down and spotted a few of these VC guys we’d been talking to about Akamai.” A few minutes later, however, they realized the opportunity at hand. “One of these guys came by and asked us how things were going,” said Seelig. “And we were able to honestly say, ‘Great, we actually just came from one of our data centers.’”
Picture the Internet as a system of highways, one designed for the high-speed travel of data, or packets, around the globe.
It’s a system that worked well early on, when the Internet served only a small community of users. But as traffic and the size of files began to increase exponentially, it became clear that the architecture of the Internet—a network of networks with no central point of command—wasn’t built to withstand it.
By the mid-1990s, its major throughways were often congested and chaotic. For home users of the Internet, it was commonplace to type an address into the Internet’s browser and wait an average of twenty seconds while the dial-up connection chirped, only to arrive at a frozen screen and the message “The server is busy…Please try again later.” That’s because the average request for information over the Internet had to make approximately nineteen hops before reaching its final destination.
Originally based on a combination of many algorithms created by Lewin and Leighton, Akamai’s debut service, called FreeFlow, was able to process real-time information on traffic conditions and use it to trick the Internet’s routers into sending data requests to Akamai servers at the edges of the Internet, closer to the end users. FreeFlow did two things to speed up the delivery of content and make Web sites feel faster. First, it adjusted BGP—the protocol used to make routing decisions on the Internet—to work almost in real time to direct data requests around the hot spots or traffic jams. Second, it cached, or stored, files on the servers in its own vast network that were closest to users.
With FreeFlow, Akamai created its own private path across the public Internet. Built entirely on software, it was a virtual road, one that took advantage of the Internet’s architecture without relying on it. This enabled Akamai to offer its customers a number of benefits, chiefly fault tolerance. If one of Akamai’s servers went down, the others in the network automatically picked up its workload until its service was restored. Another
benefit to Akamai was ease of installation. To “Akamaize” their Web site, content providers followed relatively simple, step-by-step instructions to launch the FreeFlow software, which tagged the content or pages to be served by Akamai with an Akamai Resource Locator (ARL), Akamai’s version of a URL, or Web address.
The value of FreeFlow went beyond the speed and efficiency of content delivery. Another benefit was maintenance. Customers didn’t have to install or maintain any hardware, a major draw for the engineers who would otherwise have to spend time, money, and energy on finding a good fix when their sites reached peak traffic. Akamai also offered what it called “Proof of Performance,” the guarantee that if FreeFlow failed to deliver a customer’s content at any time, or failed to deliver it faster than the customer’s own servers, Akamai would issue a refund.
No Better Time Page 10