ronutz
← Todos los fabricantes

Vendor lineage

Akamai Technologies

Answered a question Tim Berners-Lee posed about congestion, and the answer became most of how the web is delivered.

Akamai was incorporated on 20 August 1998 by Tom Leighton, an MIT professor of applied mathematics, and Danny Lewin, his graduate student, with Jonathan Seelig, Preetish Nijhawan and Randall Kaplan. It began as a response to a challenge Tim Berners-Lee had put to MIT: the web was going to get congested, and somebody should work out what to do about it.

The profile below covers the founding, consistent hashing, the 1999 listing, 11 September 2001, and the move from delivery into security and compute.

Two details sit outside that account. The intellectual property was licensed from MIT rather than owned outright at the start, which is the ordinary arrangement for a university spin-out and a real constraint on an early company. And the security business was assembled as much by purchase as by extension - Prolexic for denial-of-service scrubbing in 2014, later Guardicore for segmentation - on top of capacity that already existed for delivery.

Founding stories

1998

Akamai Technologies

Cambridge, Massachusetts · Founders: Tom Leighton, Danny Lewin, Jonathan Seelig, Preetish Nijhawan, Randall Kaplan

The problem came from down the hall. Tim Berners-Lee, then at MIT, described what was being called the World Wide Wait: a popular page collapsed under its own audience, because every request for it travelled to one server. Tom Leighton, who led the theoretical computer science group at MIT's Laboratory for Computer Science, and his graduate student Danny Lewin spent two years on it as a mathematics problem rather than an engineering one. Lewin's answer became his master's thesis. They entered the MIT $50,000 entrepreneurship competition in 1997 and did not win it, which turned out not to matter, and incorporated on 20 August 1998. The name is Hawaiian for clever.

Founder list per company histories; sources beyond Leighton and Lewin vary in who they name and in what order. Lewin's degree year appears as both SM '97 and SM '98 in MIT's own publications.

The timeline

  1. Consistent hashing

    Lewin's thesis work, developed with Leighton, on distributing requests across a changing set of servers so that adding or removing one moves as few keys as possible. It won MIT's award for best master's thesis presentation the following year.

  2. Incorporation

    Founded 20 August in Cambridge, after a business plan competition entry that did not win but attracted venture funding.

  3. March Madness, then the IPO

    The first public proof was the US college basketball tournament: sixty-four teams, a few days, and an audience that arrived all at once. The company listed on NASDAQ on 29 October, raising around $234M. The shares opened at $26 and closed their first day above $145.

  4. The crash, and 11 September

    The stock fell from over $300 to under $5 as the dot-com collapse ran its course. On 10 September the executives sat late deciding who to lay off. The next morning Danny Lewin was killed aboard American Airlines Flight 11, aged 31; he is generally held to have been the first person killed in the attacks. The same day, the network he had designed carried news sites through the heaviest traffic they had ever seen, and the remaining staff spent it completing emergency integrations for publishers whose own servers had failed.

    Flight, age and the emergency integration work per Akamai's own memorial page, the Marconi Society citation and contemporary CNN reporting. Lewin, a former Israeli special forces officer, is widely reported to have attempted to stop the hijacking; that account rests on flight-recorder analysis and is not established.

  5. Speedera

    The acquisition of its closest competitor also settled the patent litigation between them, which is a common shape in infrastructure: buying a rival and buying the end of a lawsuit at the same time.

  6. National Inventors Hall of Fame

    Leighton and Lewin inducted together, sixteen years after Lewin's death, for the algorithms behind content delivery.

Flagship products and solutions

  • FreeFlow, and what became EdgeSuiteThe original service: a publisher rewrote the URLs of its heavy objects so they resolved to Akamai, and the images and video came from a server near the reader instead of from the origin.
  • Edge delivery platformCaching and routing across a network that reached tens of thousands of servers inside carrier and ISP networks rather than in a handful of large data centres - the architectural choice that distinguishes it from later entrants.
  • Kona and App & API ProtectorWeb application firewalling and DDoS mitigation delivered from the same edge, on the reasoning that a network already terminating the connection is the natural place to inspect it.
  • Guardicore segmentationAcquired in 2021, moving the company into east-west traffic inside the data centre - a different problem from the one it was founded on.
  • Linode and connected computeAcquired in 2022, adding general-purpose cloud compute to a delivery network, and putting the company in a market it had previously sat in front of rather than in.

Key innovations

  • Consistent hashingThe foundational contribution, and it long outgrew the company. Distribute keys around a ring so that adding or removing a server moves only a small fraction of them, rather than reshuffling everything. It is now standard in distributed caches, sharded databases and load balancers built by people who have never thought about content delivery. A master's thesis became infrastructure.
  • Putting servers inside other people's networksRather than build a few large data centres, Akamai placed machines deep inside carrier and ISP networks. That is harder commercially than technically - it requires thousands of separate relationships - and it is why the network is close to users in places a centralised design cannot reach.
  • Mapping as a live systemDeciding which edge server should answer a given request is a continuous measurement problem rather than a static configuration. The routing decision responds to congestion and failure as they happen, which is the part that is genuinely difficult to copy.
  • Delivery and defence as one networkCapacity built to absorb legitimate traffic surges also absorbs illegitimate ones. The company that solved the flash crowd found it had built the natural place to stop a denial-of-service attack, and the security business followed from the delivery business rather than being bolted onto it.

Main markets

Media and software distribution first - the customers who feel a traffic spike as an outage - then commerce, banking and government, and increasingly enterprise security. Its early customer list read Yahoo, Apple, CNN and Microsoft, which is a fair description of who had a scaling problem in 1999.

It now competes against later entrants with different architectures: cloud providers bundling delivery with their own platforms, and edge companies built after the problems Akamai solved had become assumptions. Its distinguishing position remains depth of placement inside carrier networks rather than breadth of feature.

Analyst standing

  • Consistently placed among the leaders in content delivery and increasingly in web application and API protection, which reflects a company that turned its infrastructure into a security position rather than acquiring one.
  • The longer measure is that the algorithms it was founded on are taught as general computer science rather than as company technology, and the traffic patterns it was built for - everybody arriving at once - are now the normal condition of the web rather than an exception.
Akamai: company history ↗
From the company
Sources