Business Requirements of Data Platform

    ACID
    1. Atomic Transactions. Completely succeeds or completely fails.
    2. Consistency. Data is always in a valid state.
    3. Isolation. One transaction completes before another reads same data.
    4. Durability. Committed transactions are not lost.

    Other
    5. Responsiveness.
        Reading: Should normally take less than 100ms.
        Writing: Should normally take less than 1000ms.
    6. Scalability. Performance should not degrade as number of users and data increases.
    7. Security. You should be able to secure records down to the individual field and value level.
    8. Real time. Can you provide real time updates on changes to clients for collaboration.
    9. Maintainability. The database and servers should be cheap to operate and maintain long term.
    10. Sustainable. The database vendor should be likely to remain in business long term.
    11. Analytics. Data should be available for business analysis.

Where Flat SQL Databases have problems.

    Scalability and maintainability. Scaling up SQL databases with sharding and replication is risky and expensive.

Where NoSQL Databases have problems.

    No Joins. You could implement them yourselves, but every join would at least double latency.
    Consistency. Usually there is a flexible data model and often no mechanism to modify the existing schema.