I would generally expect maxRows=0 to be faster than maxRows=-1, though it's up to the database to optimize this case... it's possible PostgreSQL is getting confused. It's also possible there's that LabKey is no longer translating that 0 correctly (for historical reasons, one layer in the server treats 0 as "all rows"). Your theory about row count is possible too. What are the timings with maxRows=1 and maxRows=-1?