Shasta: Interactive Reporting At Scale Manoharan et al., SIGMOD 2016
You have vast database schemas with hundreds of tables, applications that need to combine OLTP and OLAP functionality, queries that may join 50 or more tables across disparate data sources, oh, and the user is waiting, so you’d better deliver the results online with low latency.
It sounds like a recipe for disaster, yet this is exactly the situation that Google faced with many of its business systems, especially it seems with their advertising campaign management system. Business logic and data transformation logic was becoming tangled bottlenecking development, queries were way too large to be expressed gracefully in SQL (especially when considering the dynamic aspects), and traditional techniques to speed up queries such as maintaining materialized views either increased the cost of writes too much, or gave unacceptably stale data.
Read more at The Morning Paper