Can we have your queries, not just a invalid bench without data ?
Also note, not tuning a SQL engine and comparing with MongoDB which doesn't require real tuning is not valid, you need to tune the MySQL/PostgreSQL buffer for reading & writing performance, especially on a such AWS instance which has many RAM.
PostgreSQL isn't benched and is far far more better than MySQL & PostgreSQL in terms of features, robustness and security (yes PostgreSQL has IPv6, enabled SSL by default without being crazy & CRAM authentication).
Another point, there is no transactions with NoSQL, especially mongodb, your data can be totally broken, and the upgrade scheme after using a NoSQL during years, which moving data scheme can make your apps crazy, because you have to handle all cases, or suffer from crazy bugs.
We also compared PostgreSQL 9.6 with MongoDB 3.2 to bench new huge application we developped for production (~15 SQL hard relational tables with big constraints) and the performance/CPU cost is 1.5 lower using PostgreSQL mixed relational & JSON backend than MongoDB.
Last, a reality we see in our production, MongoDB is not safe, it doesn't ensure data is correctly written, and some data can be lost on huge workload whereas relational data really ensure data is here. The mongodb replication is not really safe, it's the major risk factor.
For further actions, you may consider blocking this person and/or reporting abuse
We're a place where coders share, stay up-to-date and grow their careers.
Can we have your queries, not just a invalid bench without data ?
Also note, not tuning a SQL engine and comparing with MongoDB which doesn't require real tuning is not valid, you need to tune the MySQL/PostgreSQL buffer for reading & writing performance, especially on a such AWS instance which has many RAM.
PostgreSQL isn't benched and is far far more better than MySQL & PostgreSQL in terms of features, robustness and security (yes PostgreSQL has IPv6, enabled SSL by default without being crazy & CRAM authentication).
PostgreSQL json backend is faster than MongoDB for unrelational data, see enterprisedb.com/node/3441
Another point, there is no transactions with NoSQL, especially mongodb, your data can be totally broken, and the upgrade scheme after using a NoSQL during years, which moving data scheme can make your apps crazy, because you have to handle all cases, or suffer from crazy bugs.
We also compared PostgreSQL 9.6 with MongoDB 3.2 to bench new huge application we developped for production (~15 SQL hard relational tables with big constraints) and the performance/CPU cost is 1.5 lower using PostgreSQL mixed relational & JSON backend than MongoDB.
Last, a reality we see in our production, MongoDB is not safe, it doesn't ensure data is correctly written, and some data can be lost on huge workload whereas relational data really ensure data is here. The mongodb replication is not really safe, it's the major risk factor.