So good question - but the short is answer is there are like a million different things:
let's go with the primary factors to look at:
1) Your machine size/speed --- Alteryx runs in memory - RAM is critical for all processes - but especially critical for anything in-memory.
2) Your query size ---- Running on canvas (input data vs In-DB) - for ODBC operations means the results of your query are all retrieved into local memory. 100,000,000 row query --- it's now being sent directly to memory. For In-DB operations - push down queries are performed at the DB level but results are not retrieved.
3) With that you have a combo of network speed (how fast can you connect to your db, is their a VPN, etc) and speeds of your underlying DB (how fast can your DB process your query).
4) Driver conflicts --- is this the right driver for your db? for your machine? for Alteryx?
5) Query plan --- is your Query semi-optimized?
Those are the main buckets. Best question I've read this morning.
