Your blog/website being popular, more visitors come, you would find your website is slow and the servers are not able to handle that much of requests. However, there are some steps can optimize your website quickly. This tricks applied to your website/blog running on Drupal or WordPress or Zend Framework or any PHP Framework ...... etc.

1. Benchmarking your web server

To optimize your web server performance, firstly, we have to perform a proper assessment. Here, I am going to use "Apache Benchmark". Using this tools to query a specific page of your website and measure the response. To do so, you can execute the following command line from the server you are testing since it has low CPU and RAM footprint.

The only parameter need to play around is "c", since "n" is the number that needs to be large enough to provde stable results. The aims of this test is to test the website with various "c" values. The "c" value should start with small number and increase it until the values of "Requests per second" starts dropping.

2. Reduce memory footprint until paging stops

From the above result, we can see the requests per second drop suddenly. It is bceause there are too much requests which overload the memory and the system start using the paging file. The similar effects happen in Windows or Mac OSX when there are too many applications opened.

By execute the "free -m" command, we can see how much of swap file is used. Here, the server has 2145MB of swap file which is captured when the Apache Benchmark was run with largest "c" value.

Here, we are going to redo the test above with "c" values that are close to the one where we see drop on performance. And here are going to check the paging file after each request:

From the above result, we can see the rate of request drops significantly when there are 250 concurrent requests.

3. Configurate the max number of connections

By default, both Apache and MySQL are configurated to accept 150 connections. For most PHP applications (e.g., Drupal) will open one connection per thread. In order to know how much memory required to handle one request, you can execute TOP command:

By monitoring the result above, we can estimate how much memory is used by Apache, by averaging the value of RES (resident non-swapped physical memory) used by the processes. Here the average value of RES is about 62MB. 

Also, you can have a look the amount of free memory availble when the server is not under load:

From above, we know that there are 2720060KB (2656MB) free memory available.

Based on above value, we can use a simple calculation to find out the optimal numbers of concurrent users:

We can concluded that the server starts melting when there are more than 43 concurrent users. And here, we can reduce the number of clients from 150 to 40 in both Apache & MySQL configuration.

Once the above changes are adjusted, we can re-run the benchmark again:

As you may notice, it seems like that the Apache able to answer more than 40 concurrent request?? Actually, the Apache only process 40 concurrent requests. Any request beyond the max concurrent number are put in a queue and he client may experience some delay.

4. Performance of MySQL depends on RAM

In fact, MySQL does not consume much CPU. MySQL will consume 10 - 25% of CPU if the server well configurated. However, MySQL is very RAM consuming. It is common to see MySQL use close to 50% of the RAM available even MySQL is having its caching and threads optimized.

To optimize MySQL performance, the rule is to give as much RAM as possible to MySQL.

5. Performance of Apache depends on CPU

On Drupal, WordPress or others website based on PHP Framework, Apache spends most of its time to execute PHP code. Apache divide the available CPU across each one to answer request. Hence, CPU is the key resources of Apache.

Tags: Server Setting/Optimization