Netflix’s Cloud Migration

Netflix has easily become one of the top streaming services worldwide. It kickstarted DVDs into extinction and is one of the most prominent players in its field. Before it came to be such a powerhouse, like many businesses, Netflix faced struggles with disasters and outages.

Starting in 2008, Netflix was hit with a shock that changed the way they looked at and handled their databases. The IT department assumed that if they attempted to make their systems perfect, then developers would no longer need to worry about future failures. Netflix was running on high-end and costly IBM P-series hardware, Oracle database. The assumptions proved false when a SAN hardware failure resulted in a two day outage. Across management, questions were raised. Eventually, the company came up with the declaration that availability was concerning applications. As the application was the matter of the situation, Netflix’s people realized that there was no need for expensive hardware, so they would use a more cost friendly cloud infrastructure. It was more than the question of whether this strategy was going to push them forward rather than was it simply working.

A year later, with Netflix’s growth, they found themselves in need for more data center capacity. The rapid increase for Netflix did not allow them to spend the time to estimate how much capacity and where to store all the data. The future logistics were unpredictable since previous data were based on the shipping of DVDs. With the advancement of IT, scaling is now based on systems of engagement and how many customers they have. Previously with Netflix’s DVD business, there would be only one or two engagements per week. Customers were only browsing and watching movies a few times a week since they were limited to how quickly they would receive the product. Many traditional enterprises are similar in this way of customer interaction. Looking at the now widely used streaming business of Netflix, people are binge watching episodes of television daily. With the streaming service, there is more that goes into it like progression logging. If a customer were to stop halfway through whatever they were watching, the system would know where they got to. Quality of service in total is more advanced. Interactions during streaming is also amplified versus in the DVD business. Netflix calculated a thousandfold increase of traffic in the datacenter with the streaming service. With the switch to streaming was growing quickly, Netflix was faced with the issue of building data centers just as fast.

Netflix was faced with two options: recruit expensive, world class data center operations and guess how much data they would need to build centers before it was needed or use Amazon Web Services, created by one of Netflix’s biggest competitors. With the latter route, Netflix would save tremendous amounts of money which could be used for development and video content. Netflix decided to go with option number two, allowing for extra cash to be invested back into the company.

Later in 2009, mitigating risks was one of the top priorities. Competition, capacity, business and publicity were four aspects most focused and explored on. They learned more about AWS and learned it was separate from Amazon Prime, Netflix’s competitor. Netflix conducted experiments on capacity to see what worked, how quickly systems could be grabbed, and which datacenter would be deployed. Additionally, Netflix signed up for a business license agreement with AWS so they would not have to run the service on a click through license on a credit card. Fast forward to April of 2010, New York Times wrote an article on the unity of AWS and Netflix. They were the first to join cloud computing at the time and the article caught a lot of publicity.

After transitioning to cloud computing via AWS, Netflix found they could receive capacity on demand allowing for backlogs to be shut down. As more time passed, Netflix came to the point where if they did not transition all mission critical applications to AWS they would not survive. A hard deadline was created to migrate to the cloud to further concentrate the focus of Netflix’s priorities. Overall, the data migration sequence started with the simplest API sequences, then the simplest web pages, and finally the rest of the web pages and API’s one by one. The gradual migration was accomplished by some of the site traffic directed to the old data centers, while AWS was serving the others when possible.

Like any other business utilizing the cloud, Netflix also participated in a vigorous data backup strategy to ensure their success. The Netflix we see and use today exists because they transitioned to cloud computing. With the development of technology in recent times, any company seeking to migrate to the cloud has an easier and simpler progression.