Funding
Funding not disclosed
Founders
Product
Problem
In distributed systems, transient failures, rate limits, and the need for resumability introduce complexity in application development, often requiring developers to implement custom error handling and retry mechanisms. This leads to increased development time, code duplication, and potential inconsistencies in failure recovery across different parts of the system.
Solution
Stealth Rocket's Dispatch provides a decorator-based solution that simplifies the addition of automatic retries, execution resumability, and rate limiting to Python and Go functions. By wrapping functions with the `@dispatch.function` decorator, developers can ensure reliable execution without extensive code modifications. Dispatch automatically retries failed functions, restores state from the last run, and resumes execution where it left off, preventing duplicate operations. The system dynamically adapts execution rates to stay within defined limits, using an algorithm similar to TCP's congestion control to optimize traffic flow and prevent overload.
Target Audience
The primary users are Python and Go developers building distributed systems, microservices, or applications that interact with unreliable APIs and require robust failure handling and rate management.
Features
- `@dispatch.function` decorator for Python and Go functions to enable automatic retries, resumability, and rate limiting.
- Automatic retry mechanism that handles transient failures and restores state from the last successful execution point.
- Dynamic rate limiting that adapts execution rates on the fly to avoid overloading systems or exceeding API limits.
- AIMD-based algorithm for optimal rate limiting, similar to TCP's congestion control.
- CLI tool for easy setup and deployment, including example projects.
- No vendor lock-in: removing the decorator allows the code to function as before.