Hi @anirudh286, How are you doing today?

As per my understanding, The HA option in Databricks picks instances from multiple zones right from the start, not just when one zone is unavailable. This ensures your cluster is always highly available but could lead to cross-zone data transfer costs. If you're concerned about those costs, you might want to carefully consider whether HA is necessary for your project or look for ways to reduce data movement between zones.

Just give a try and let me know.

Regards,

Brahma