...
The only prerequisite is to fill out an online request for a NAAN on behalf of your organization. There is no charge to obtain a NAAN and all memory organizations are welcome. Within a day or two you should receive an email containing a NAAN for your organization's exclusive use. Meanwhile consider the following.
...
...
It's like serving ordinary URLs. Incoming URL strings get mapped to content that you return, and if your resolver redirects ARKs to those URLs, you're all set. If you're dealing directly with incoming ARK strings, you can map (convert) them to a form your server handles (eg, map them to URLs on arrival). In this second case, your server is acting as a local resolver.
If you choose to run your own ARK infrastructure, you get complete autonomy at the expense of maintaining a server/resolver. On the one hand, you might run all custom infrastructure – including content management, web hosting, minting (generating unique identifier strings), and running your own server/resolver. That infrastructure could be very simple, such as server configured to map incoming ARK-based URLs to server file pathnames. When you request your NAAN you will be asked to supply the base URL of your local server or resolver.
At the other extreme, you might work with a vendor that supplies all the infrastructure so that, for example, you can focus on creating content. Hybrid solutions are also common, such as just taking your current web server arrangement and just adding an identifier management piece (eg, the API/UI provided by ezid.cdlib.org, which partners with n2t.net).
You will also want to think about whether to advertise (release, publish, disseminate) your ARKs based at your resolver or at n2t.net. You might choose the former for branding or the latter for stability. Resolving your ARKs through n2t.net is always possible, regardless of how you advertise them (this is a side-effect of obtaining a NAAN).
...
Content negotiation for metadata is a kludgy software technique for requesting alternate formats of an object, such as the PDF or RTF form of an HTML file. Although not designed for it, historic "content negotiation" was contorted kludged (twisted) in certain contexts to request metadata under the startling assumption that formats often used to hold metadata are in fact metadata and will never be objects in their own right. Unlike inflections, "content negotiation for metadata" doesn't work at all for objects represented in those formats (the list of which is growing and known only by private agreement), nor is it easy enough to be used directly by most human users.
...