Scalable Selenium Cluster: Up & Running
summary
Claims a production cluster with >5,000 browsers, 1.5M sessions/day, ~4,000 requests/sec and ~800 Mbps traffic built using Dockerized browsers and a lightweight proxy called Ggr server that uses session ID magic to remain stateless. Browser processes run in isolated containers via Selenoid, cutting memory and dependency bloat compared to selenium-server, exposing the standard Selenium API, supporting auth, autoscaling across data centers, central logging/metrics, and a 10k+ Docker Hub pull count for its images.
AI-written summary · about
chapters (8)
- 0:00Introduction and scale
- 2:37Problems with Selenium Grid
- 6:43Introducing selenoid
- 10:17Demo: installing selenoid
- 14:50Building a scalable cluster
- 20:33Introducing ggr
- 24:57Demo: configuring ggr
- 28:58Monitoring and conclusion
related talks
44mReliable. Scalable. fast. Running 1000000 Android tests with Selenoid
41m1001 Reasons to Use Ggr in Your Selenium Installation
45mRubber Cluster for Dynamic Selenium Grid (Opening Keynote)
40mSelenium Grid
44mZalenium: Use A Disposable And Flexible Selenium Grid Infrastructure
25mEfficiency Matters: Mastering Dockerized Scalable Grid for UI Tests
44mSelenoid: Ditch Selenium Server! Part 2
45mSelenium in Kubernetes: Lessons Learned
