Peter Bissmire

17-08-01, rev. 18-08-03

Canonicalisation

...is currently a major buzzword in parts of the SEO (search engine optimisation) community, who use this term for the ability to reach, by entering domain.tld in the browser address bar, the same site as is reached by entering www.domain.tld. Let's examine this mythology.

Consider first the word canonical. This means in accordance with the rules. In the domain name system (DNS), a canonical name constructed in accordance with the rules, e.g. hostname.mydomain.tld, points, again in accordance with the rules, to an IP address in an A record. The host may have other names but these are all aliases and point, in CNAME records, to the canonical name.

One of the definitions put forward for canonicalisation is, "www.mydomain.tld and mydomain.tld must resolve to the same URL".
This displays a complete lack of understanding of how things work on the Internet. Both www.mydomain.tld and mydomain.tld are domain names, mydomain.tld being a "naked" domain name and www.mydomain.tld being an FQDN (fully qualified domain name). Neither will get you directly to a host (e.g. Web server); this requires the intermediary of DNS. This will tell you the name servers for mydomain.tld and their IP addresses and these will, in turn, tell you the IP address for the host www.mydomain.tld. Once you have the IP address you can contact the host.
DNS resolves domain names to IP addresses (and the other way round). To talk of domain names resolving to URLs is thus absolute nonsense.

It has even been suggested by some canonicalisers that the Web server should redirect requests to mydomain.tld to www.mydomain.tld.
To be thus redirected you must first find the server and the redirect must be its default response. If you have found the server, you are already there, so why not make its default response the Web-site for www.mydomain.tld? That's exactly how this server is configured. This does not, however, make the server automatically respond to requests for mydomain.tld, they won't reach it unless special, dare I say subversive, measures have been taken in the DNS.

This is, in fact, the job of the Web browser. When, on the basis of the information provided, it ends up with an NS (name server) DNS response then, by convention and in the absence of an explicit host-name prefix, it prepends www. to form an FQDN. If this, too, fails then it can refer the user entry to a search engine. This has been the case ever since the days when Netscape was almost the only kid on the block.

This "canonicalisation" was already dealt with decades ago. For a discussion of the proper use of the term, see Wikipedia.