Serverless computing refers to an architectural pattern where server-side code is run on-demand by cloud providers, who also handle server resource allocation and operations. Of course, there is a server involved on the provider’s side, but administrative functions to manage that server such as capacity planning, configuration, or management of containers are handled behind-the-scenes, allowing the application developers to focus on business logic. This makes for highly elastic and scalable systems and can reduce development, testing, and iteration time due to reduced overhead.
Function as a Service (FaaS) describes a model of serverless computing where services are decomposed into modular functions and deployed to a serverless platform. These functions are executed only when called and are typically stateless. Despite the benefits of elasticity and modularity that FaaS offers, it has drawbacks as well. Taking disaggregation of functionality to an extreme means that behavior that formerly required a method call now may require a network call to another function, increasing latency and making larger-scale operations inefficient.
Cloudburst is a stateful FaaS platform built to combine the power of low-latency mutable state and communication in Python with the elasticity and scalability allowed by serverless architecture. Johann Schleier-Smith is an entrepreneur and engineer who is currently a board member of Sama. Johann was formerly the founder and CTO of if(we), a social network and incubator. He is the co-author of the paper “Cloudburst: Stateful Functions as a Service,” and joins the show today to talk about how Cloudburst addresses the drawbacks of current FaaS models, and what’s next for serverless computing.
Sponsorship inquiries: email@example.com