This tool can be used to benchmark various Database performance under "no-sql" usage conditions.
For the full explanation, check https://eadalabs.com/benchmarking-mysql-postgres-and-mongodb-performance-in-nosql-mode/
The database are used in "no-sql" mode, i.e. without any join between table. The objective of the test is to insert as fast a possible into the table, and read-back as fast as possible.
Two table configuration are defined:
-
light table:
- First column: index (int, primary key)
- Second Column: integer
- Third Column: integer
-
large table:
- First column: index (int, primary key)
- Second Column: var char of 200 characters
- Third to 10th Column: integer
-
Write
- Insert to a table with 10 concurrent threads without transaction
- Insert to a table with 10 concurrent insrrt thread using a transaction flushed every minute (only one thread for the transaction flush)
- Insert to a table with 10 concurrent connections without transaction
-
Read
- Read the whole table data with one single thread
- Read the table data with 2 theads: the first threads read the first half of the values, and the second thread reads the second half.
-
Mysql
make write-innodb
make read-innodb
make write-myisam
make read-myisam
-
Postgress
make write-postgres
make read-postgres