> On the other hand, a project that rarely updates information, such as an analytics application where data doesn't change after it has been collected, will benefit more from a nonrelational database.<p>SQL data warehouses are relational but have far better analytic query performance than databases like MongoDB. They generally use column rather than row storage. The difference is (a) column stores don't read unreferenced columns and (b) columns compress extremely well. On large datasets databases like ClickHouse tend to process analytic queries orders of magnitude faster than MongoDB given the same machine resources.