Home/ Questions/Q 7682097
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-05-31T18:29:20+00:00

As I do not trust many benchmarks out there, because they are not actually

  • 0

As I do not trust many benchmarks out there, because they are not actually handling I/O like database connections, just sending “Hello World” to user and recieving enormous throughputs. I am not building any scalable things at the moment but plan to do soon. I need better results and make sure it’s realistic as much as possible.

I execute the following query at both scripts: “SELECT * FROM table” which returns 200 queries. The sample results are:

(Both MySQL, NodeJS anda Apache Servers are restarted before each test)

1000 requests, 100 concurrency:  Apache: 564.20 req/s   NodeJS: 102.61 req/s
1000 requests, 200 concurrency:  Apache: 393.75 req/s   NodeJS: 105.12 req/s
1000 requests, 400 concurrency:  Apache: 42.75 req/s    NodeJS: 105.07 req/s
1000 requests, 1000 concurrency  Apache: 16.5 req/s     NodeJS: 119.53 req/s

However, I do not trust these results, Because the query is always the same which is not likely in the real-time applications. How can I test the both suites myself? Would changing SQL query to an random insertion or random selection would help? Some caching in PHP-MySQL case helps if I do not restart the servers.

The code I used to bencmark:
NodeJS:

var mysql = require("mysql");
var os = require('os');
var MySQLPool = require("mysql-pool").MySQLPool;
var client = new MySQLPool({
  poolSize: 40,
  user:     'root',
  password: '123',
  database: 'test'
});

var express = require('express'), routes = require('./routes');
var app = module.exports = express.createServer();

app.get('/tables', function(req,res){
    client.query(
      'SELECT * FROM sample s',
      function selectCb(err, results, fields) {
        if (err) {
          throw err;
        }
        res.send(results);
      }
    );
});

var cluster = require("cluster");
if (cluster.isMaster) {
    for (var i = 0; i < os.cpus().length * 2; i++) {
        var worker = cluster.fork();
    }
} else {
    app.listen(3000);
}

PHP

$host = "localhost";
$username = "root";
$password = "123";
$db_name = "test";

$db=mysql_connect($host, $username, $password) or die('Could not connect');
mysql_select_db($db_name, $db) or die('no db');
$sth = mysql_query("SELECT * FROM sample s");
$rows = array();
while($r = mysql_fetch_assoc($sth)) {
    $rows[] = $r;
}
echo json_encode($rows);
?>

I know there are other approaches like Redis and Memcached to improve the performance, but I am interested in raw MySQL performance.

Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    I’d recommend an insertion over any sort of selection — the former can’t be cached by nature, while the latter takes extra effort to prevent the same.

    It’s also worth timing a super-simple SELECT query. Replace “phoneinfo” with any table you have available;

    SELECT COUNT(anyFieldName) FROM phoneinfo WHERE 1 = -1;

    If you run both of those, and log the timing from 100+ rapid-fire page loads, then you should get a much better sense of what the comparison is.

    • 0