Consistency is the “C” in the ACID properties required for validity in the event of errors, network partitions, and power failures. The four ACID properties are Atomicity, Consistency, Isolation, and Durability.
Are your database schemas stable?
If your database schemas are unlikely to change significantly over time, and you want most fields to have consistent types from record to record, then SQL databases would be a good choice for you. Otherwise, NoSQL databases, some of which don’t even support schemas, might be better for your application. There are exceptions, however. For example, Rockset allows for SQL queries without imposing a fixed schema or consistent types on the data it imports.
Geographic distribution of users
When your database users are all over the world, the speed of light imposes a lower limit on database latency for the remote users unless you provide additional servers in their regions. Some databases allow for distributed read-write servers; others offer distributed read-only servers, with all writes forced to go through a single master server. Geographic distribution makes the trade-off between consistency and latency even harder.
