Companies either had to duct-tape compute engines together, such as fast key-value stores, analytical in-memory engines, streaming engines, machine learning libraries, and notebooks, or use expensive scale-up solutions, like SAP HANA or Oracle Exadata …