Give Your Apache a Performance Boost

You can improve the performance of the popular Web Server, Apache, by integrating mod_cache and Deflate modules. The first enables faster web page access whereas the second increases download speeds

Saturday, August 02, 2008

Caching of web pages is regularly done to speed up access. Let's suppose you have an Apache web server deployed locally. By deploying mod_cache you can get the server to regularly update the contents of frequently visited sites. This way the visitor has access to the latest version of a page as soon as he visits it and Apache is spared the need to check for it real time. This saves a lot of bandwidth for enterprises. Apache 2.x also provides the option of integrating 'deflate' module, that allows one to improve the downloading speed by compressing the downloaded packets.

How the server uses these modules is actually pretty simple. It first checks if the URL is present in its cache or not. It the URL is present and the version has not expired, then it serves the page directly from the cache else the page is loaded from the respective server. In such a scenario (while the page is being served directly from the server), Apache applies filters to the page to check if it contains cacheable content or not. In case it's cacheable, then the web page is cached by mod_cache for future use. It makes use of two other modules: mod_disk_cache and mod_mem_ cache. Let's see how you can implement this in real life scenario. Before we start, make sure that mod_cache module is installed. For this, run the following command, '/usr/sbin/httpd -M.'

This will display all modules that are installed in the Apache module set. Check if cache module is present or not. If not, then download 'Apache cache module' rpm package from rpmfind.net and install it on your web server using the following command:

# rpm –ivh <cache module package name>

After you have installed the module, it's time to configure the configuration file of Apache. This is to tell Apache that it should load the cache module and use it. Uncomment the following lines in httpd.conf file, in case the line is marked as a comment; this would allow Apache to load the cache module. The file can be found at the location: /etc/httpd/ conf/httpd.conf.

LoadModule cache_module modules/mod_cache.so
LoadModule disk_cache_module modules/mod_disk_cache.so
LoadModule file_cache_module modules/mod_mem_cache.so

Apart from the mod_cache module the server needs two other modules: mod_disk_cache and mod_mem_ cache. Mod_disk_cache is used to for disk based caching or storing and mod_mem_cache is used for memory based caching. The second caching option should be used when the content is generated locally. Now place the following lines into httpd.conf:

<IfModule mod_disk_cache.c>
CacheRoot /var/www/cache
CacheEnable disk /
CacheDirLevels 5
CacheDirLength 3
</IfModule>
<IfModule mod_mem_cache.c>
CacheEnable mem /
MCacheSize 4096
MCacheMaxObjectCount 100
MCacheMinObjectSize 1
MCacheMaxObjectSize 2048
</IfModule>

In the first module we have defined where the content is to be cached, ie at '/var/www/cache'. Then we enabled the caching option; defined the directory level, ie the depth of the directory; and finally the number of characters in a subdirectory name. In the second module, ie mod_mem_cache, we first enabled the caching option then defined the size of cache inside the memory. Next, we defined the maximum objects that can be stored in the memory and the min and max size of the object. This has to be defined so that the memory is correctly used else caching a single file of 1 GB in memory can't be considered. Finally, you enabled all the caching options in Apache. For cases where the page is not cached and is served directly from the server, you can implement the Deflate module. It is a data compression algorithm that uses the combination of LZ77 algorithm and Huffman coding.

This will help you in case your content is being downloaded directly from the server. The content is first compressed then sent to clients, which makes data transfer faster over the network. Before we start, check whether the Deflate module is installed or not. For checking, use the same method as used for mod_cache. If the module is not installed, download the Deflate module rpm from rpmfind.net and install it in the same way as you installed other modules. Once the module has been installed, copy the following line inside the 'Location' tag in the httpd.conf file:

AddOutputFilterByType DEFLATE text/html text/plain text/xml

This will enable Apache to compress all HTML and XML files. But if you want to enable compression for other types of data such as images and scripts, into Apache, then type the following lines as well:

<Location />
AddOutputFilterByType DEFLATE text/plain
AddOutputFilterByType DEFLATE text/xml
AddOutputFilterByType DEFLATE application/xhtml+xml
AddOutputFilterByType DEFLATE text/css
AddOutputFilterByType DEFLATE application/xml
AddOutputFilterByType DEFLATE image/svg+xml
AddOutputFilterByType DEFLATE application/rss+xml
AddOutputFilterByType DEFLATE application/atom_xml
AddOutputFilterByType DEFLATE application/x-javascript
AddOutputFilterByType DEFLATE application/x-httpd-php
AddOutputFilterByType DEFLATE application/x-httpd-fastphp
AddOutputFilterByType DEFLATE application/x-httpd-eruby
AddOutputFilterByType DEFLATE text/html
<Location>

Finally you have a full set of solutions in place for serving web pages faster to the clients. Apart from this, there are directives to optimize mod_cache. For example CacheMaxFileSize and CacheMinFileSize help you in defining the minimum and maximum file size that needs to be cached on the disk. The syntax would be (numbers specified in bytes):

CacheMaxFileSize 20000
CacheMinFileSize 2000

MCacheMaxStreamingBuffer is another such directive for mod_mem_ cache. This can be used when there is some streaming data to be cached. As you cant predict the size of the data being streamed, you inadvertently end up having falling short on cache. Using this directive you can specify the maximum size of the buffer for streaming data. So the moment the buffer gets full, it discards any further data; hence you are spared of having your cache space getting saturated. This is useful as the streamed data doesn't have Content-Length header which tells the size of the data. Syntax for implemenging the buffer is, 'MCacheMaxStreamingBuffer 10000.'

Page(s)   1