How does Selenium Grid work and how do you use it?

Selenium 4 Grid uses a hub-and-node model with a router, session map, distributor and node processes. The router accepts HTTP commands, the distributor queues sessions, and nodes advertise browser capabilities. Grid does not make tests parallel by itself; the runner decides how many sessions to request.

ChromeOptions options = new ChromeOptions();
options.setBrowserVersion("stable");
options.setPlatformName("linux");

WebDriver driver = new RemoteWebDriver(
    new URL("http://grid.example.com:4444"),
    options);

try {
    driver.get("https://example.com");
    Assert.assertTrue(driver.getTitle().contains("Example"));
} finally {
    driver.quit();
}

• Capabilities determine which node can satisfy a session request
• If no node matches, the session stays queued or fails after the configured timeout
• A node can run multiple browser slots depending on CPU and memory
• The session map tracks active sessions so commands reach the right node
• Grid nodes are usually containers in modern deployments, often via Docker Selenium
• Video, VNC and logs can be enabled per node for debugging

Size the Grid from observed concurrency and test duration, not from the number of test methods. Each browser slot consumes significant memory. A Grid with too few slots creates long queues that look like slow tests, and a Grid with too many slots on one host causes browser crashes and timeouts.