Network-transparent messaging combined with a shared nothing architecture enables actors to run concurrently without risking data races. Once applications outgrow a single machine, CAF enables users to scale out seamlessly by distributing actors across the network.

Actors naturally thrive in an ever-changing environment and scale horizontally just as easily as vertically. This makes actors the perfect fit for building applications that grow and shrink dynamically with user demand.

Leveraging C++ enables us to tightly control memory layout as well as actor scheduling. As a result, CAF actors consist of only a few hundred bytes and the messaging layer reclaims memory as soon as possible.

Whether building large-scale applications that live in the cloud, or IoT application that run on resource-constrained devices, CAF users can rely on a low memory footprint and fast execution thanks to a minimal runtime overhead.

Distributed systems have many sources of errors. Writing software for the real world means to anticipate partial failures and to gracefully respond to hardware, software and network errors. CAF cannot make crashes and bugs go away, but it equips you with the tools to manage them.

Actor-based design organizes applications into sub-systems with failover strategies that allow forming supervision trees. These techniques proved themselves in practice time and time again for building reliable, fault tolerant systems.