Hi Jody,
The POC test run than I'm doing does both reads and writes. Without getting into specifics, the application loops through a workload of inserting data into a master-detail set of tables - similar to Orders and Line items. There are then a set of 7 queries that run based on PK/-FK joins - most are two to three tables. There is one query that does some heavy expressions in SQL, but one table. There are no temp tables created as part of the plans and no joins between row and column tables.
The main issue is, with SQL Server, I get a much more detailed explain plan that includes missing index information. There is also a great body of community knowledge on how to identify bad plans and things like unexpected table scans. For SAP HANA, I see very little out there on how to diagnose explain plans. For row tables, I can make better guesses about what indexes are needed to cover my main queries, but given the columnar nature of COLUMN tables - none of the DB vendors offer help. Some vendors like Redshift, don't even have a CREATE INDEX command!
The ACM paper gives the impression, that with SAP HANA, creating indexes on column tables should be discouraged, but doesn't help in calling out then they are useful. So, I'm calling out for help.
Regards,
Bill