Testing / Selenium Interview questions
What is the difference between Selenium Grid's Hub-Node model and the new Router-Distributor model?
Selenium 3's Grid used a single centralized Hub that both accepted incoming test requests and directly tracked every registered Node's capacity - one component doing both routing and capacity management.
| Hub-Node (Grid 3) | Router-Distributor (Grid 4) |
| Single Hub handles routing and capacity tracking together | Router and Distributor are separate, independently scalable components |
| Hub is a single point of failure/bottleneck | Components can be scaled or run independently |
| No built-in session queueing | Distributor includes a session queue for when Nodes are full |
Splitting routing (Router) from capacity/assignment decisions (Distributor) is what makes Grid 4 more resilient to a single component becoming a bottleneck, and it's also why Grid 4 can run in a fully distributed mode across containers or Kubernetes pods, with each component scaled to its own actual load rather than everything funneling through one Hub process.
More Related questions...