What are typical middleware choices for handling asynchronous calls in Redux?

When managing asynchronous side effects (such as API calls, debouncing, and web sockets) in Redux, several middleware libraries have evolved:

### 1. Redux Toolkit Query (RTK Query) - Modern Recommendation
The current standard provided out of the box with Redux Toolkit (@reduxjs/toolkit). RTK Query eliminates the need to hand-craft thunks, reducers, and loading/error states for server data fetching, providing automatic caching, invalidation, and polling.

### 2. Redux Thunk
The most common and lightweight approach. A thunk is an action creator that returns a function (dispatch, getState) instead of an action object, allowing asynchronous promises to be awaited before dispatching resolved actions:

export const fetchUser = (id) => async (dispatch) => {
  dispatch({ type: 'user/pending' });
  try {
    const res = await api.getUser(id);
    dispatch({ type: 'user/fulfilled', payload: res.data });
  } catch (err) {
    dispatch({ type: 'user/rejected', error: err.message });
  }
};

### 3. Redux Saga
Uses ES6 Generator functions (yield takeEvery, yield call, yield put) to orchestrate complex asynchronous workflows. Ideal for applications requiring cancellation, debouncing, concurrent worker pools, or complex background synchronization.

### 4. Redux Observable
Built on RxJS Observables, treating actions as streams. Highly powerful for real-time finance feeds, live chat, or complex event-driven pipelines requiring RxJS operators like switchMap, debounceTime, and mergeMap.