[MUSIC]

Now that we have a solid understanding of the general domains and

associated tools in the world of big data,

let's begin to look at a specific use case of these tools applied to real problems.

Let's take a look at Netflix,

one of the largest video content providers in the world.

Did you know that they are responsible for

35% of all the traffic on North American networks?

Netflix handles millions of concurrent users on a daily basis.

They must satisfy numerous business requirements and

service level agreement to keep their network latencies low,

their content recommendations accurate, and their services online.

All of their efforts need to scale to millions of users, so they invested in

many of the technologies that we've discussed throughout this course, and

have even developed many of their own in-house.

Most of the tools that they've created have been open sourced and

are in heavy use across other companies dealing with similar issues,

companies like Yelp and IBM, Squarespace, Yahoo, and many more.

So before we dive into how the issues were solved,

let's think a bit about what their issues are specifically.

The first issue I see is storage.

There are a ton of movies and shows on Netflix, right?

Well, they need to store massive amounts of video data in a way that can be

delivered quickly to a user in any part of the world where the company operates.

Now to address this problem, Netflix has had to invest in major content delivery

networks, or CDNs, database storage systems, caching systems, and

messaging pipelines to handle passing the content around their internal networks and

to their users.

Next, that sure is a lot of active users.

Processing all those incoming user interactions across many

different clients, from laptops to mobile phones, to Apple TVs, cable boxes,

video game systems?

They have to provide recommendations in real time for which content to watch next.

Other non-user based actions like monitoring and

logging need to be aggregated across all these clients as well.

Creating many streams of data that need to be processed in the back end, and

acted upon by other systems in order to keep the systems online and running well.

To address this problem, Netflix relies on a variety of stream processing services,

searching systems, and machine learning tools.

The next obvious problem is running the infrastructure needed to serve

all the traffic and users.

Netflix has to scale out thousands of machines across many geographic regions,

using Amazon Web Services, Google Cloud platform, and

self-hosted services to process, store, search,

and learn from the data produced by clients and backend systems.

To address this problem,

Netflix uses several of the infrastructure tools we have discussed.

Let's look more specifically into how they do that, right after this quick break.