Discover definitions of serverless and learn about how a key business scenario drove Netflix to a cloud-native serverless architecture.
- [Instructor] As we get started thinking about serverless architectures, I find it helpful that we have a common definition. So, here's how I define serverless services. These are services that can be used for compute, data, and other types of capabilities in your application. For these services, you would write the code for some situations, other situations you would simply call functions. The service itself would be provided by your provider, which in many cases is the public cloud, but could also be a private cloud or a third-party service.
Notice you're working at the level of a function, not at the level of a container, or a server, or virtual machine. This is an important distinction that we'll be exploring throughout this course. Your functions would execute on event triggers. Understanding the new world of data expressed as events rather than data that comes over in a structured and batch form is really critical to understanding some of the new application architectures that serverless enables.
And the third component is you pay for these services only when executed, and the vendor, or whosever hosting, will autoscale these services for you. To get us started thinking about core serverless architectures, let's consider the Netflix Story. As one of the lead architects at a critical juncture, Adrian Cockcroft, who at the time of this animation was working at Netflix. He currently works at Amazon.
You can see that when he was working at Netflix there was this critical point indicated on this graph, when customer behavior was changing from ordering a DVD and performing that as a batch operation, a datapoint in a relational database if you will, to getting the information or the movie from Netflix as stream data. You'll notice that Adrian says, only .1% of the customers were starting to stream, but as the Netflix employees were forecasting the amount of data coming in through those event streams, it was growing exponentially.
And as he says, at that critical juncture they had an important decision to make. Should they either A, buy tremendous data center capacity to meet that anticipated demand, or should they move to a new paradigm, and Adrian says, we knew at that point we had to change our paradigm to a serverless cloud architecture.
- Defining serverless architecture
- Serverless services
- Serverless functions
- Cloud-native serverless architecture
- Serverless architecture for big data and machine learning
- Emergent serverless architecture