In the diagrams earlier in this series, we used http://davidjanes.myopenid.com as David's "identity URL" -- basically, the analogous thing to a username in OpenID. David is also using MyOpenID as his "identity server" -- the site he will log into.

There is no requirement in OpenID that your "identity URL" be on your "identity server".

OpenID provides for a layer of indirection between your identity URL and your identity server. When logging in, the "consumer" (the role BlogMatrix.com is playing in the diagrams below) looks at the identity URL, reads the page, looks for a special link identifying the identity server and if found, uses that.

Let's spell that out with an example:

• David's identity URL is now http://www.davidjanes.com
• David types http://www.davidjanes.com into BlogMatrix's login page
• BlogMatrix reads the page at URL "http://www.davidjanes.com" and sees in the HTML header:
  
  <link rel="openid.server" href="http://www.myopenid.com/server" />
  <link rel="openid.delegate" href="http://davidjanes.myopenid.com/" />
  <meta http-equiv="X-XRDS-Location" content="http://davidjanes.myopenid.com/xrds" />
  
  
• BlogMatrix knows how to understand all that stuff to use MyOpenID as the server

Note that the particulars for all these fields will change if you're not using MyOpenID, but it'll be something very simular.