Skip to content

Latest commit

 

History

History
180 lines (132 loc) · 6.48 KB

README.md

File metadata and controls

180 lines (132 loc) · 6.48 KB

Overview

The sql-benchmark provides simple benchmark of several database drivers in several languages. Benchmarks are intended to be simple so that they can be implemented easily for several languages and SQL databases. Some of the goals are:

  • Evaluate the performance of a database driver,
  • Compare the performance of different languages when connecting to a database,
  • Have a rough comparison on simple SQL queries on different databases.

Languages

Databases

Three SQL databases are supported:

  • SQLite,
  • MySQL/MariaDB,
  • PostgreSQL

Before running the SQL benchmark of MySQL/MariaDB and PostgreSQL, you must create the sqlbench database and give access to the sqlbench user.

The SQLite database is created automatically.

MySQL/MariaDB setup

  1. Create the 'sqlbench' database in MySQL/MariaDB
mysql -u root
mysql> create database sqlbench;
  1. Create the 'sqlbench' user:
mysql> create user 'sqlbench'@'localhost' identified by 'sqlbench';
  1. Give the access rights:
mysql> grant select, insert, update, delete,
       create, drop, create temporary tables, execute,
       show view on sqlbench.* to sqlbench@'localhost';
mysql> flush privileges;

Postgresql setup

To create manually the database, you can proceed to the following steps:

  1. Create the 'sqlbench' user and configure the password (enter 'sqlbench' for the password or update the configuration sqlbench.properties file):
sudo -u postgres createuser sqlbench --pwprompt
  1. Create the 'sqlbench' database in Postgresql
sudo -u postgres createdb -O sqlbench sqlbench

Running

The script run-all.sh can be used to run all the benchmark and produce the results. Before running it, make sure you have built the Ada and Java benchmark programs as well as the Ada aggregator tool. To build, run the following commands.

cd ado
./configure
make
cd ../java
mvn compile assembly:single
cd ../tools
./configure
make
cd ..

Then, simply run the script:

./run-all.sh

Results

Time Memory SQLite MySQL PostgreSQL

CONNECT; SELECT 1; CLOSE

sqlite mysql postgresql
Ada 23.68 us 311.0 us 5.541 ms
Java 187.7 us 895.8 us 10.80 ms
Python 42.00 us 398.5 us 6.071 ms

DO 1

sqlite mysql postgresql
Ada 19.04 us
Java 53.72 us
Python 33.03 us

DROP table; CREATE table

sqlite mysql postgresql
Ada 216.9 us 496.2 ms 42.64 ms
Java 1.698 ms 504.9 ms 45.03 ms
Python 122.5 ms 498.6 ms 41.24 ms

INSERT INTO table

sqlite mysql postgresql
Ada 73.59 us 118.5 us 160.6 us
Java 96.20 us 241.4 us 119.3 us
Python 3.684 us 290.7 us 119.1 us

SELECT * FROM table LIMIT 1

sqlite mysql postgresql
Ada 25.63 us 75.60 us 105.5 us
Java 5.096 us 97.38 us 105.6 us
Python 3.457 us 47.31 us 148.7 us

SELECT * FROM table LIMIT 10

sqlite mysql postgresql
Ada 27.60 us 61.30 us 99.43 us
Java 5.115 us 101.7 us 92.56 us
Python 8.766 us 51.76 us 128.0 us

SELECT * FROM table LIMIT 100

sqlite mysql postgresql
Ada 37.41 us 131.1 us 161.6 us
Java 15.53 us 174.5 us 223.8 us
Python 60.43 us 102.0 us 236.6 us

SELECT * FROM table LIMIT 500

sqlite mysql postgresql
Ada 78.14 us 328.6 us 305.7 us
Java 62.12 us 462.2 us 616.2 us
Python 297.8 us 300.2 us 462.2 us

SELECT * FROM table LIMIT 1000

sqlite mysql postgresql
Ada 132.0 us 544.9 us 456.6 us
Java 121.8 us 728.2 us 871.6 us
Python 605.6 us 551.6 us 730.3 us

SELECT 1

sqlite mysql postgresql
Ada 9.501 us 35.60 us 87.55 us
Java 5.629 us 70.61 us 104.4 us
Python 1.530 us 78.13 us 89.66 us