The biggest issue is that Druid doesn't really support JOINS well. Druid has a lot of complexity by denormalizing the data. I'd look at StarRocks which does JOINS, faster and also integrate with data lake open formats like Apache Iceberg and Apache Hudi.