Good, Cheap, and Fast, Scaling Your Selenium Grid in the Cloud
summary
Company runs ~10,000 browser tests per day and migrates a static, undersized Selenium Grid onto AWS using EC2/Lambda/DynamoDB, Datadog metrics and a Lambda autoscaler to add/remove nodes based on queue length, shaving cost from ~$4,000 to ~$1,800 (55%). Adding Spotinst for spot instances drops average node cost to ~$41 and integrating a custom Amp Liam proxy that load balances across multiple hubs and forwards sessions to Sauce Labs yields about $230/month total (≈95% reduction).
AI-written summary · about
chapters (18)
- 0:00Intro
- 1:06What is Amplify Education?
- 2:09Amplify's First Selenium Grid
- 3:34First Foray Into The Cloud
- 8:51Handing Hurg Sessions
- 9:49Instaling Latest Selenium
- 10:33Instaling Latest Browsers
- 11:29Why Do We Autoscale?
- 14:06Adding Metrics
- 16:46Acting on Metrics
- 20:52What is the Soot Market?
- 26:03Why Not The Grid Hub?
- 26:59What is Amplium?
- 28:29How Does The Load Balancing Work?
- 29:28How Does The Queueing Work?
- 30:33How Does The SauceLabs Integration Work?
- 31:51Next Steps For Amplium
- 33:01Network Diagram
related talks
40mSelenium Grid
40mScaling Selenium to infinity using AWS Lambda
46mServerless - how to speed up tests over 300 times?
44mServerless – Automated Tests of web apps in the Cloud world
09mTest Time Execution! Why it Can't be Ignored?
43mInfinite Space for Your Tests
36mRunning Automation tests twice as faster
44mZalenium: Use A Disposable And Flexible Selenium Grid Infrastructure