Modern applications often suffer from accidental complexity where resiliency code is mixed with business logic, creating a 'smoothie architecture'.
Monoliths are often simpler than microservices because database transactions eliminate the need to handle partial failures, a major source of complexity.
Event-driven architectures introduce problems like unclear APIs, scattered business logic, poor error handling, and low visibility into system state.
Events create high design-time coupling because changing a producer's event can unknowingly break multiple downstream consumers, similar to changing a global variable.
Durable execution is an abstraction where code is guaranteed to run to completion by resurrecting its execution state after process crashes or infrastructure failures.
Durable execution encapsulates complexity by providing durable local variables, long-running functions, infinite retries, and simplified asynchronous communication.
A long-running subscription workflow can be implemented in just a few lines of pure business logic using durable execution, eliminating boilerplate code.
The state of a durable execution is recovered by replaying a stored history of external events, which deterministically reconstructs the application's state without serialization.
A complete durable execution platform requires features beyond the core abstraction, including uniqueness constraints, signaling, querying, indexing, and versioning.
Temporal is a scalable, open-source platform that provides a production-ready implementation of the durable execution model for building resilient applications.
29:32Maxim Fateev
37:51Geetha Anne
57:38Bernd Ruecker
24:25Florian Lenz
32:46Wei Hu
50:19Mihaela-Roxana Ghidersa
28:41Wei Hu
18:07Maxim Fateev
Jobs that call for the skills explored in this talk.
Peter Park System GmbH
München, Germany
CONTIAMO GMBH
Berlin, Germany
Datadog
Paris, France
DL Remote
Düsseldorf, Germany
DL Remote
Dortmund, Germany
DL Remote
Dresden, Germany