RunAt.dev Platform

Details on RunAt.dev's platform


RunAt.dev allows scheduling API calls to run at any time in the future.

A priority for our platform is initiating all requests as soon possible to their scheduled time, with as little latency as possible.

Millisecond-level Timing Precision

Most scheduling systems run on a cron-like interval, where a task can be run every day, hour, minute, etc. With a system like this, it's impossible to have sub-minute task scheduling and very hard to dictate exactly when tasks run since they're interval based.

With RunAt.dev, it's possible to schedule an exact time a task runs, down to a millisecond, perfect for unique, one-off scheduled tasks or user-defined scheduling.

RunAt.dev achieves this by not using a cron-like interval system, and instead using a unique alarm system that "wakes up" your task to be run exactly at the time you set. This alarm system allows us to provide more granular scheduling, which is how we can offer millisecond precision.

At the Edge

RunAt.dev lives at the edge, meaning our platform runs on many servers, all over the world, close to you. This is crucial to minimize latency from our servers to yours.

Tasks are both stored and run in a server geographically closest to where they're created, to minimize latency when sending a request.

All necessary components to run a task live at the edge. All of the timing, storage, and networking, the entire critical path to running a task, happen in a server close to you.

Server Isolation

RunAt.dev prevents your tasks when slowing down or having latency increase when our platform is busy by isolating each task from one another.

Each scheduled task can be thought of as existing in its own server at the edge, created on demand. When your scheduled task runs, this server immediately executes it, never waiting on anything external, to minimize latency.

At least once delivery

RunAt.dev guarantees at least once delivery of each scheduled request, so you can depend on never missing a request. In very rare cases, duplicate requests may be made.