Dart Isolates: True Parallelism Without Shared Memory

Why it exists

Most user interfaces, including Flutter's, run on a single thread. If you perform a long-running task like complex math or parsing a large file on this main thread, the UI freezes because it can't draw new frames or respond to user input. Isolates were created to solve this by moving heavy work to a different thread.

The mental model

Think of an isolate as a separate workshop with its own tools and materials (memory). Workers can't just grab things from each other's benches; they must send messages back and forth through a pneumatic tube system (Ports). This "share-nothing" concurrency model prevents data corruption and race conditions by its very structure, removing the need for manual locking.

How it works

When you spawn an isolate, the Dart VM creates a new thread with its own memory heap and event loop. You provide it a function to execute. To communicate back to the main isolate, you use SendPort and ReceivePort objects. You send a message through the SendPort, and the receiving isolate listens for it on its ReceivePort. The data sent is copied, not shared, ensuring total isolation.

When to use it

Use isolates for any CPU-intensive task that could block the main thread for more than a few milliseconds. The canonical use case is in Flutter, to prevent "jank" or unresponsiveness while processing data. Examples include parsing and decoding exceptionally large JSON objects, image processing, cryptography, or running complex algorithms.

When not to use it

Do not use isolates for I/O-bound operations like network requests; async/await is designed for that and is more lightweight. A major limitation is that multiple isolates are not supported in Flutter web applications, which operate in a single-threaded browser environment. For very small computations, the overhead of creating an isolate and copying data may also be slower than just running the code on the main thread.

One canonical example

A Flutter app needs to parse a 50MB JSON file. Parsing it on the main thread would freeze the UI for several seconds. The correct approach is to use Isolate.run() to perform the JSON decoding on a background thread. The isolate takes the raw string data, parses it into Dart objects, and sends the final result back to the main thread, which can then safely update the UI without ever dropping a frame.