ingesting high volume small size files in azure databricks

32 Views Asked by At

Databricks documentation says that autoloader can ingest millions of files per second from cloud storage using event notification mode. However, when I enable event notification mode, speed is at best reaching 300 files pers second. All files are quite small 10-50KB in size.

Have tried both directory listing and file notifications. Both show same speeds. I suspect file notification mode is not working properly. Even though I can see that it has setup the event grid and queue storage in azure automatically after I provided required access / permissions

0

There are 0 best solutions below