A Totally Ordered, High Throughput, Pub/Sub System with Disaggregated Memory

Publish-subscribe messaging systems sit at the core of today’s data-driven clouds. These systems decouple the services that produce and consume messages, providing an asynchronous communication abstraction that, in theory, can be scaled to any number of producers and consumers. In practice, however, existing systems are often presented with a choice between scalability and desirable global properties ike durability and strong ordering guarantees.

Embarcadero presents an unprecedented combination of total ordering, high throughput, low latency, and efficient fault tolerance facilitated by physically disaggregated memory. Embarcadero also allows for near-instantaneous elasticity and churn of broker nodes, providing better scalability and availability as long as a single broker remains in the rack. Overall, Embarcadero advances the state-of-the-art in shared log messaging systems, offering a compelling balance of strong consistency and high performance.


Jaewan Hong, Erika Hunhoff, Tony Hong, Marcos Aguilera, Emmanuel Amaro, Vincent Liu, Ion Stoica