Load from HDFS to Oracle

ubethke

When inserting data from HDFS to Oracle or another supported database such as Greenplum, does the data flow through the DSS server (memory)?

Mattsco

Hi Ubethke,

No, hopefully you won't load all your data into the memory of the DSS server.

The data is "streamed" by chunk from HDFS to Oracle.

You should check this learn article for more details.

Matt

ubethke

Based on the article it is streamed from HDFS to DSS server to Oracle. Correct?
If so then I would argue that this is not efficient as it requires an unnecessary roundtrip of the data through the DSS server.
Your thoughts?
Thanks
Uli

Mattsco

Correct!
I agree but it's not always technically possible for us to avoid this roundtrip.