WWWhat’s in a Name?

December 15, 2011 by . 7 comments

A recent question asked:

What does wwws mean? Is it to do with Security?

I’ve seen a couple of sites that use wwws in their domain, for example: wwws.mint.com and wwws.whitehouse.gov. I’m sure it has something to do with an extra layer of security. What does it mean and what is it used for? Why do only a handful of sites use it?

Summary

The www and wwws prefixes don’t affect the browser’s choice of HTTP or HTTPS protocol.

A few organisations use wwws. to suggest that HTTPS is supported and arrange redirection so that users don’t have to type the https:// protocol specifier.

Currently, few organisations think this is worthwhile. There is a trend to drop prefixes like www..

The parts of a URL

Before we look at what wwws means, it might be useful to look at where it fits in, as far as web-page addresses (URLs) are concerned:

A URL such as http://www.example.com/fruit/apple.html consists of the following parts:

http://
The protocol specifier – There are many protocol specifiers, each one tells the web-browser what network protocol it should use to fetch the information referred to by the URL. Some examples include http:// https:// ftp:// file:// but I’ll just look at http:// in this article.
www.example.com
The hostname – this is an example of  a fully qualified domain name (FQDN). The example.com part tells us the URL is an address for data from a particular business company. The www part names a particular computer operated by (or for) that company. So this part tells the web-browser which server-computer to send it’s request to.
/fruit/apple.html
The path – This is passed by the web-browser to the server and used by the server that receives the request to find the information requested. Usually it is the last part of a filename path that refers to a file that contains the exact information to be returned unaltered. Often it is the last part of a filename path that represents a program or script that is to be run at the server and which produces the information to be returned – the program might do this by referring to a database. We can’t tell for sure from the path but if the path ends with a “filename extension” of .html it is likely to be a file to be returned unaltered, if it ends .php or .asp (or many other filename extensions) it is almost certain to be a program or script to be run at the server. Only the server knows for sure

More complicated URLs are also possible. For example: http://username:[email protected]:81/fruit/apple.php?q=russet&lang=en. Here, 81 is a port-number which is like a sub-address at the server. The server uses this port-number to decide which service (which primary program running on the server) should be given the request. If you don’t specify a port-number in the URL, the browser uses port 80 for URLs with a http:// protocol specifier. The extra-part at the end (after the question mark) is additional information that is eventually passed to a secondary program or script that uses it to decide what actions are taken or what information is returned.

So – what does wwws mean?

Nothing. It is only a convention.

The prefix www is non-functional, it has no effect on the protocol used. However I believe Tim Berners-Lee, the father of the World-Wide-Web, used it as a naming convention to indicate an HTTP server (i.e. a World Wide Web server). This helped to distinguish the server name from servers for other information-sharing protocols – E.g. some corporations had servers named ftp.example.com, www.example.com, mail.example.com (or maybe even gopher.example.com and wais.example.com). These prefixes were convenient for network administrators who wished to be able to locate these different services on separate servers. Nowadays other means of doing this have arisen. It is arguable the hostname prefixes were not convenient to end users. Most web browsers have tried to remove this burden from users by automatically adding http://www. and .com when the user enters single word attempts at a URL.

What is it used for?

To suggest that the server may support privacy of data in transit and certification of server identity.

A few people have taken to using wwws as a naming convention for HTTPS servers, i.e. servers that support HTTP over Secure-Sockets-Layer (SSL).

Since the name has no effect on protocols, you can run a HTTPS service on a server named example.com or foo.example.com or anything else.

The proponents of the wwws convention say that wwws is shorter and easier to type than https:// and they can therefore configure their servers to redirect a HTTP request for a wwws hostname to an HTTPS URL as a convenience for users who explicitly want to use HTTPS instead of HTTP but who find typing https:// too much of a chore.

There are other ways to provide links to a secure version of a site. Most business that need security (shops, banks) will automatically redirect users to HTTPS pages without requiring users to type a special prefix to the hostname.

Why do only a handful of sites use it?

Since it has no standardized functional purpose, almost nobody finds it a useful convention. The trend is to drop prefixes such as www.

Try visiting these URLs:

  • http://www.superuser.com
  • https://www.superuser.com

Then check which URL shows up in the address bar of your browser.

So why not get rid of the www prefix?

Perhaps we should. However there are times when the example.com in http://example.com/fruit/apple.php is not enough.

Using different tools for different jobs

Sometime we have a complicated website with complicated web-pages, some pieces of a page might need to be fetched from a database using complicated code. For example the text of a Question and associated Answers in superuser.com. Other parts of the page might not need any processing – for example a picture associated with a particular person. The latter are sometimes referred to a s “static content” (and the complicated part as “dynamic content”).

For very popular sites (like superuser.com) it can be a bit wasteful (and therefore slower and more expensive) to use a sophisticated and versatile server program to serve up simple static content. So there is a desire to separate the two jobs (dynamic bits, static bits) amongst two different primary server programs (i.e. “web-server” software) that are each optimised for a specialised task – or at least where one is optimised for static content. It may also be useful to split the job between two computers. Therefore we need two names, one for the main computer and one for the computer responsible for static content. So we might choose www.example.com and static.example.com. A visitor to the site would never actually see the “static.example.com” because that would only be used inside a web-page (HTML) for references that the web-browser uses to get images etc that are part of a page.

But not so fast! We could equally well chose just example.com and static.example.com. There isn’t an absolute requirement to use www. Even if there were, there would be no reason not to use, say, web.example.com instead of www.example.com. Though something more meaningful would be even better (blog.example.com?)

So it turns out this isn’t a good reason to add www to the start of example.com.

Spreading the load

Big successful sites have too many visitors every minute for a single computer to cope with all their requests. Therefore big sites use a “farm” of many computers for the job. But we don’t want some users to have to try lots of different server names until they find one that isn’t busy. We want one name for all the servers!

One way to handle this is sometimes called “round-robin DNS”. When you use your web-browser and clink on a link (or bookmark) to http://www.example.com/fruit/, the browser asks a directory service to look up the numeric address for www.example.com (because the Internet is actually based on numeric addresses like 10.232.17.105). This directory service is usually one called DNS. A DNS server can be set up to associate not just one numeric address with example.com but a whole list of them. Moreover it can be configured to return that list in a different order each time, influencing different people’s web-browsers to use different numeric addresses and hence reach different computers. Thus the load is spread across lots of computers.

Now some say that to achieve this you can’t associate multiple numeric addresses with example.com. You have to use name aliases (in DNS jargon they are implemented using somthing called CNAMEs) to say that www1.example.com is just an alias for www.example.com and www42.example.com is also an alias for www.example.com. A long time ago I used to administer DNS for a large corporation and I am unaware of why you can’t associate CNAMES with just plain old example.com. Probably I have forgotten and the other people are right.

However, even so, there is nothing magic about the name www. It could as well be any other name (web, blog, info, whatever)

Conclusion

So www and wwws don’t really mean much, they’re just traditional (some say clutter).

Filed under Browsers

7 Comments

Subscribe to comments with RSS.

  • I should also point out that the ‘www’ subdomain can cause problems if your site’s security certificate wasn’t issued for subdomains of your primary domain. Chrome users, for instance, will see a page with a large red warning indicating that the site’s certificate is not valid for the the current domain and suggesting that they avoid the site.

    One can, however, purchase certificates that are valid for subdomains (up to one level).

  • It is a shame there is no question linking/aliasing between the SE’s as this would be good to have on a couple of SE sites. Oh well.

  • nhinkle says:

    Christopher: you can always answer a related question on another site with a link to this blog post or to the original question on Super User.

  • Shane Madden says:

    Multiple A records work just fine on a domain root, as would pointing www1 etc to it via CNAMEs (noting, of course, that using www1 without giving those names to clients gains you no load balancing). What doesn’t work is pointing your root as a CNAME record to somewhere else (like making a CNAME record at example.com pointing to http://www.example.com).

  • soandos says:

    There are a few times when the www. is required to access the site, or the default server supplies a redirect to the www server. Has messed me up a few times.

  • Bryan says:

    One thing to note is that fact that many web browsers have now stopped displaying the protocol part of the URL in the location bar. Whilst not a massive issue, this might have influenced a few sites into somehow adding the ‘s’ back, if this is the case, then they are greatly misleading users, potentially luring them into a false sense of security. Personally, I don’t see this browser trend as progress, as I much prefer to see the full URL (but I am a geek rather than a user).

  • Svick says:

    Try visiting these URLs:

    https://www.superuser.com

    My Firefox gives a security warning about this one, saying that it uses invalid certificate. If I understand it correctly, there certificate is valid for *.stackexchange.com, which does not include superuser.com.

  • Leave a comment

    Log in
    with Stack Exchange
    or