Amazon Web Services, which is worth $60 billion, is today a major player in the cloud infrastructure market. It was just a startup in 2008 and still struggling to meet growing demand for its cloud servers. Amazon EC2 was launched in beta last week 15 years ago. AWS was a major selling point for startups at that time, offering unlimited compute power.
EC2 was the first attempt to sell elastic computing at large scale. This meant that server resources could scale up when you needed them, and disappear when you didn't. In 2008, Jeff Bezos stated in a sales presentation to startups that lightning can strike and that if you are not prepared, it will be a huge regret. It's hard to accept if lightning strikes and you aren't prepared. You don't want to be too prepared for the worst. That's why [AWS] can help with this difficult situation.
Animoto was a startup customer that Animoto helped to scale from 25,000 users to 250,000 in just four days. This happened shortly after the company's South by Southwest Facebook app launched in 2008.
Animoto, an app that was targeted at consumers at the time, allowed users to upload their photos and convert them into videos with backing music. Although it may seem simple today, the product was still a major step in the right direction back then. Each video required a lot of computing resources. This product was a pioneering example of Web 2.0 user-generated content. It also represented the marriage of mobile computing and the cloud, which we now take as a given.
Animoto was founded in 2006 and chose AWS. However, the company discovered that running its own infrastructure was more risky due to the changing nature of its customers. It would have required huge capital investments to spin up its servers. Animoto started out that way, but then turned its attention to AWS as it was still building before it received initial funding. Brad Jefferson, the cofounder and CEO of the company, explained.
In order to prove the concept, we built our own servers. We started to build our own servers and gained more traction as a proof of concept. Jefferson explained to me that we should prepare for success.
Although it may seem easy to choose AWS based on what we know now, the company was actually putting its fate in the hands a relatively unknown concept in 2007.
It's quite interesting to see the progress made by AWS and EC2, but it was a risky venture back in those days. We were speaking to an e-commerce firm about running our infrastructure. They tried to convince us that their servers would be dynamic, and they did. Jefferson said that although it seemed obvious now, it was still a risky proposition for companies like ours to place a bet on them.
Animoto needed to trust that AWS could deliver what it claimed. It also had to spend six month redesigning its software so it would run on Amazons cloud. Jefferson looked at the numbers and made the right decision. Animoto's business model was available for free for a 30-second video, $5 for longer clips, and $30 for a full year. He tried to model how much resources his company would require to make his model work. It became really hard so he and his cofounders decided to put their money on AWS. They hoped it would work when there was a surge in usage.
South by Southwest's next year was a test when it launched a Facebook application. This led to an increase in demand and pushed the limits of AWSs capabilities. After the app was launched, interest surged and Amazon had to scramble to find the right resources to keep Animoto running.
Today, Amazons VP for EC2 is Dave Brown. He was an engineer on that team in 2008. They had been using 50 to 100 instances per day for the past month. Their usage reached 400 on Tuesday, 900 on Wednesday, and 3,400 instances by Friday morning. AWS was able provide the resources necessary to support this surge in demand. Its usage reached 5000 instances, before it dropped back down. This proved that elastic computing can actually work.
Jefferson stated that his company was not reliant on EC2's marketing. The company was constantly on the phone with AWS executives to ensure that their service would not be affected by this growing demand. The most important thing was: can you get more servers? We need more servers. Although I don't know how they managed it, they were able get us to where we needed to go. He said that we were able then to get past the spike and things naturally settled down.
Animoto's story of staying online was a major selling point. Amazon was the first company to invest in it, aside from friends and family. The company raised $30 million in total, the last funding being in 2011. The company is now more of a B2B business, allowing marketing departments to easily create videos.
Jefferson did not discuss the specific costs but he said that it was difficult for companies to afford servers that were often idle. Jefferson claims that cloud computing proved to be the best model, and his company remains an AWS customer today.
Cloud computing's goal has been to offer as much computing as possible on demand, whenever you need it. However, the circumstances in this case put that idea to the test.
Today, the notion of having difficulty generating 3,400 instances seems absurd, especially considering that Amazon processes 60,000,000 instances per day. But back then, it was a significant challenge that helped startups understand that elastic computing was more than a theory.
