Spark Performance Advisor

Automatically detect performance issues in all your Spark jobs

Try Free

Detect common issues

... and not just data skew

  • Spill — when you're out of execution memory
  • Excessive retries — when the data is recomputed all over
  • Unused executors — when the driver is doing too much
  • Several more now, even more coming in future

One job or thousands

Track your entire workload

Got thousands of jobs? You can use Spark Advisor to automatically find performance regressions as data changes

Troubleshoot specific jobs

One job misbehaves? Run it with additional Spark options and we can probably suggest what's wrong

No cost, 5 minute setup

And no marketing calls either

  • Setup a Spark listener in 5 minutes
  • Run the job as usual
  • Find the key metrics in a simple UI
  • Receive an overall score of common problem
Go to console Documentation