One shared harness drove every system

Continuous NBBO
stock quote stream

1M rows/sec target

100B rows / ~28h

Drill-down
queries
over raw data

2 queries every hour

Interactive
aggregate queries

4 queries every
10 minutes

Shared ingestion client

Same data · Same pacing
Same client compute
Provider-specific delivery adapter

Same preparation
requirements

Same sorting / clustering keys
Same daily pre-aggregations
Platform-native implementation

Shared query driver

Same queries
Same schedule
Throughout ingestion