How Netflix lends its power to cloud computing architecture it

With the rapid development of cloud computing, the application of cloud computing in various industries and fields has become an unstoppable trend.

The biggest example of cloud computing applications today is the online video leasing service, Netflix, which uses cloud computing to implement it architecture. It's called the biggest because Netflix has a huge monthly video access rate of over 1 billion times.

The Linux Foundation Collaboration Summit was held in San Francisco recently, and Netflix's cloud system architecture director Cockcroft revealed that Netflix's Linux, FreeBSD and various open-source services are pure "native cloud computing". This can definitely become another innovation in the application of cloud computing.

Cockcroft is saying this because although there are more than 1 billion video visits per month, Netflix does not have its own data center, and in 2009 Netflix started using Amazon's AWS Cloud services, and by November 2012, Netflix has shifted all IT infrastructure to Amazon's AWS Cloud.

Specifically, Netflix's front-end services run on 500-1000 linux-based Tomcat JavaServer and Nginx Web servers, based on customer demand, with Amazon's S3 cloud storage and NoSQL Cassandra database servers ( Adopt memcached High-performance distributed Memory object caching system. All of these services are released through three of the most available cloud areas in the Amazon AWS, ensuring that users can quickly browse videos, whether they are using a mobile device or a browser.

If one of the three Amazon cloud areas fails, Netflix can still provide complete services through the remaining two cloud areas, Cockcroft said. Cockcroft said it was not a copy of Amazon's advertising words, but rather an open-source software chaos gorilla The results of a global test of AWS.

It is not easy to make a backup of the video and services in the cloud range, and Netflix now has its own cdn--open Connect and a large number of servers deployed at the local ISP node. But the really difficult question is how do you get the DNS system to point traffic to the available cloud areas in time when a cloud area fails? As we all know, DNS and APIs are very different, the design principle of DNS more rely on the Engineer manual configuration, so it is difficult to realize automation. There is now a technology plan within Netflix that the goal is not how to stop cloud computing downtime, but rather the downtime of cloud computing, rather than a sudden paralysis like a power outage, which can buy time for subsequent measures.

It should be noted that Netflix's cloud computing architecture is still at risk, such as a load-balancing failure in the December 2012 Amazon AWS East Coast Data Center, which made Netflix's services unavailable during Christmas.

Although Netflix's cloud computing architecture is not perfect, but by moving the entire IT infrastructure to the cloud, Netflix realizes the high agility and availability of IT architecture: The new feature code can be released in a few days instead of the past few months, configuring and enabling new hardware in minutes instead of the past few weeks, Netflix built the most bandwidth-hogging media platform on the public cloud and quickly grew into an internet giant that could subvert the traditional television industry, all without the success of its cloud computing architecture.

Original link: http://www.ctocio.com/ccnews/12150.html