For dedicated browser type node setups, say we want to test against Firefox, Chrome, Edge, Opera, and Safari browsers. Do we have enough resources to create Windows, Mac, and Linux platforms for all these browsers? Or do we care more about testing on different browsers instead and are somewhat ambivalent to the platform they run on?

Here is a design to support dedicated browser type nodes:

Set up each node to only create instances of one browser type. For this scenario, you would need nine VM nodes, as follows:

• Windows 10/Firefox x 5 instances
• Windows 10/Chrome x 5 instances
• Windows 10/Edge x 5 instances
• macOS/Firefox x 5 instances
• macOS/Chrome x 5 instances
• macOS/Safari x 5 instances
• Linux/Firefox x 5 instances
• Linux/Chrome x 5 instances
• Linux/Opera x 5 instances

So, in essence, although you need 10 VMs for this grid structure (1 hub, 9 nodes), you actually have 45 virtual platforms to test against. If we test on the Windows 10 node using Firefox, we can have five separate test suites running on that node at the same time.

This design allows five separate Firefox browsers to be running simultaneously on the node, since multithreading is built into the driver class. Each thread handles its own set of instructions, directed of course through the grid hub, which will not interfere with others tests running on the node.