Redirects: TYPO3 and RealURL vs mod_rewrite

Very few people know how to optimize TYPO3 web sites for performance. There is one common mistake that almost everyone does in regard to TYPO3 performance. The mistake is to let TYPO3 handle redirects.

Why are TYPO3 redirects bad?

Let's see two typical cases when redirects happen. The first one is domain redirects, the second one is a redirect using RealURL.

Domain redirects

Domain redirects happen when there are several domains on the same page and all domains except one (primary) should redirect to the primary domain. This is usually done for SEO purposes.
How is it usually handled? The administrator creates a domain record for the primary domain. Next he creates domain records for all other domains and specifies the address where to redirect them (to the primary domain).
How do this work? Let's trace the chain of actions involved into handling HTTP request for a non–primary domain:

1. Apache receives a request
2. (if RealURL is enabled) Apache uses mod_rewrite to rewrite the request to index.php
3. PHP invokes index.php from TYPO3
4. The whole TYPO3 framework initializes
5. (if RealURL is enabled) RealURL resolves a page id
6. TYPO3 finds that it should redirect
7. TYPO3 sets corresponding HTTP headers to PHP
8. PHP returns headers to Apache
9. Apache sends redirect reply to the browser

This is an approximation, of course. But there are at least these 9 steps involved. Steps 2 to 8 are useless waste of server resources. We know that domain always must be redirected. Thus invoking the whole TYPO3 for this task is a waste of resources.

RealURL redirects

Let's see what happens when RealURL redirects happen:

1. Apache receives a request
2. Apache uses mod_rewrite to rewrite the request to index.php
3. PHP invokes index.php from TYPO3
4. The whole TYPO3 framework initializes
5. RealURL finds that it should redirect to another location
6. RealURL sets corresponding HTTP headers to PHP
7. PHP returns headers to Apache
8. Apache sends redirect reply to the browser

Here we have 8 steps but it is not much better. Heavy TYPO3 is still initialized, MySQL receives queries, RealURL works and all this happens just to perform a well–known static action: redirect from one static location to another static location. It is like going to the bookshelf, taking the dictionary and searching for a known word every time when you encounter it in the text. Waste of resources.

Proper redirects

Most redirects are static. So they do not have to go through TYPO3 at all. We can use mod_rewrite to efficiently do all rewrites. Let's see how to do it properly!

Proper domain redirects

To redirect from one domain to another we can use this rewrite rule:

RewriteEngine On
RewriteCond %{HTTP_HOST} ^example.com$
RewriteRule (.*) http://www.example.com/$1 [L,R=301]

This rule redirects from example.com to www.example.com and keeps the request URI the same. There are only three steps involved in this action:

1. Apache receives a request
2. Apache uses mod_rewrite to rewrite the request
3. Apache sends redirect reply to the browser

No PHP, no TYPO3, no MySQL! Fast and efficient.

Proper path redirects

To redirect from one path to another we can use this rewrite rule:

RewriteEngine On
RewriteCond %{REQUEST_URI} ^/path/to/old/page/$
RewriteRule .* /path/to/new/page/ [L,R=301]

This rule results in the same three steps as the previous one.

Final notes

This part contain minor notes on the topic.

To slash or not to slash

When using mod_rewrite rules, it is important to check the slash in the beginning of the URL. If the rule is in the .htaccess or inside the <Directory> section of Apache configuration, there must be no forwarding slash. Both .htaccess and <Directory> are directory–related. When using mod_rewrite directives directly inside <VirtualHost>, the beginning slash is necessary because the path there is not relative.

Where to place the rules?

The best place for mod_rewrite rules is Apache configuration files. Modern systems are fast, .htaccess files are cached in memory, so there is no major decrease in performance when using .htaccess. It was a problem years ago but not now. However it is not a reason to spend even extra millisecond on parsing .htaccess. Placing mod_rewrite rules to Apache configuration files is better because saves parsing time for each request.

Why does RealURL have redirects at all?

Well, it is a long story. The feature is historical. It is convenient. And it was made when people did not realy care about performance. Now people do. So do not use RealURL redirects. You can do better with mod_rewrite!