Partitioning is a functionality that is transparent to the SQL interface. That's true for basically all RDBMS implementation.
For the SQL user a partitioned table looks and feels just as a non-partitioned table.
This in turn means: HANA, being the SQL client to Oracle here, can neither acknowledge nor ignore the partitioning feature. All it sees is a table.
Coming to the actual question: why is the data transfer so slow?
Without any measurements of the involved processing steps, there's no way to know that.
I'd start of by trying to find out what time is spend on Oracle, what time on the way between Oracle and HANA and what time in HANA.
Once you know where the time is spend, further analysis can be decided upon.
- Lars