This presentation will explore the new work in Spark 3.1 adding the concept of graceful decommissioning and how we can use this to improve Spark’s performance in both dynamic allocation and spot/preemptable instances. Together we’ll explore how Spark’s dynamic allocation has evolved over time, and why the different changes have been needed. We’ll also look at the multi-company collaboration that resulted in being able to deliver this feature and I’ll end with encouraging pointers on how to get more involved in Spark’s development.
Connect with us:
Website: https://databricks.com
Facebook: https://www.facebook.com/databricksinc
Twitter: https://twitter.com/databricks
LinkedIn: https://www.linkedin.com/company/databricks
Instagram: https://www.instagram.com/databricksinc/
0 Comments