No, the latest season of “Arrested Development” won’t fatally crash Netflix, despite comedian David Cross’s tongue-in-cheek comment that the series will melt down the company’s servers on its first weekend of streaming availability.
“No one piece of content can have that kind of impact given the size of what we are serving up at any given time,” a spokesperson wrote in an email to Slashdot.
Although “Arrested Development” struggled to survive during its three seasons on Fox (from 2003 to 2006), the series has built a significant cult following in the years following its cancelation. Netflix commissioned a fourth season as part of a broader plan to augment its streaming service with exclusive content, and will release all 13 new episodes at once on May 26.
Netflix plans to produce five new shows a year, which will bring it into direct competition for eyeballs with HBO, AMC, and other cable networks. “The goal is to become HBO faster than HBO can become us,” Netflix chief content officer Ted Sarandos told GQ back in January. The company’s opening gambit in that strategy was House of Cards, a political drama starring Kevin Spacey as a manipulative congressman, which Netflix CEO Reed Hastings termed “a great success” in terms of viewership. (However, he declined to break out actual ratings numbers—perhaps taking a cue from Amazon CEO Jeff Bezos, the Great Sphinx of Tech.)
The strategy could be paying off: Netflix has added millions of streaming subscribers, and its stock price is on the rise after a lengthy period of anemic performance. But how does the company actually handle all those users wanting their content streamed instantly to their televisions and tablets?
Like Facebook, Google, and other Internet giants, Netflix has invested quite a bit in physical infrastructure and engineers. It stores its data on Amazon’s Simple Storage Service (S3), which offers a significant degree of durability and scalability; it also provides bucket versioning, which protects data from inadvertent data loss.
“Reading and writing from S3 can be slower than writing to HDFS,” read a January note on the Netflix Tech Blog. “However, most queries and processes tend to be multi-stage MapReduce jobs, where mappers in the first stage read input data in parallel from S3, and reducers in the last stage write output data back to S3. HDFS and local storage are used for all intermediate and transient data, which reduces the performance overhead.”
Netflix’s reliance on Amazon includes the latter’s Elastic MapReduce (EMR) distribution of Apache Hadoop, an open-source framework that helps companies run massive data applications on large hardware clusters. Netflix developers rely on a number of tools within the Hadoop ecosystem, including Hive and Pig; the company also relies on “Genie,” a Hadoop-based Platform-as-a-Service (PaaS), for dynamic resource management.
Netflix contributes some of its internal work to the open-source community. Publicly available projects include Asgard, which helps manage cloud deployments via Amazon Web Services (AWS), and SimianArmy, which instigates random failures as a way to stress-test cloud infrastructure.
Outside of engineering, Hastings has done his best to imbue Netflix with a rather unique corporate culture, which is neatly summarized in a PowerPoint deck available online. “Great workplace is not espresso, lush benefits, sushi lunches, grand parties, or nice offices,” reads one representative slide. “We do some of these things, but only if they are efficient at attracting and retaining stunning colleagues.”