ganlvtech / mysql-mongodb-simple-kv-pair-benchmark

MySQL vs MongoDB simple k-v pair benchmark

MySQL vs MongoDB simple k-v pair benchmark

Prepare

Create MySQL table

CREATE TABLE `kv` (
	`k` CHAR(100) NOT NULL COLLATE 'utf8mb4_unicode_ci',
	`v` MEDIUMBLOB NULL DEFAULT NULL,
	PRIMARY KEY (`k`)
)
COLLATE='utf8mb4_unicode_ci'
ENGINE=MyISAM
;

MongoDB collection index is automatically created in mongodb_write.php with the following code.

$collection->createIndex(['k' => 1], ['unique' => true]);

Result

Item	Value
MySQL Write	25s / 20000 insert * 100000 bytes
MongoDB Write	24.8s / 20000 insert * 100000 bytes
MySQL Read	5.3s / 20000 select * 100000 bytes
MongoDB Read	14.4s / 20000 find * 100000 bytes

Item	Value
MySQL Write	20.6s / 200000 insert * 1000 bytes
MongoDB Write	36.6s / 200000 insert * 1000 bytes
MySQL Read	20.9s / 200000 select * 1000 bytes
MongoDB Read	51.6s / 200000 find * 1000 bytes

So, is RDBMS slow and NoSQL fast? Is there any problem in my test script?

I could also test using InnoDB with auto commit disabled (SET AUTOCOMMIT=0;). It got same time like MyISAM. When auto commit enabled. It costs 30 times time as the former one. 7.2s for 2000 insert * 1000 bytes.

MySQL is really fast if you use bulk insert or SET AUTOCOMMIT=0 before insert. (BEGIN TRANSACTION and COMMIT is same as autocommit=0) You cannot say that MySQL is slow than MongoDB generally. But you can say that under specific circumstances. What's more. You can optimize you MySQL database schema to get similar performance like MongoDB or even better.