Not all Distribution Networks are the same: ADN vs. CDN

I’ve misled readers, and it’s past time to straighten things out. ADNs are not the same as CDNs.

Start with definitional glosses: “ADN” generally signifies “application delivery network”; “CDN” stands for “content delivery network” or “content distribution network”. In the past, I’ve mentioned that, while CDNs complexify the task of application performance management (APM), CDNs often are a good value: a relatively inexpensive way to arrange scalability for your Website.

While this is true, it’s also sloppy, and has the potential to confuse. The problem is that many people use “CDN” as a comprehensive cover term which includes ADN. I’m accustomed to this usage, especially in business circles, and rather thoughtlessly adopted it in recent installments of  “Application Monitor“.

Dynamic vs. static

The distinction between the two is real, though, and worth understanding. Traditionally, CDN delivers static content from a geographically-dispersed collection of fileservers. Canonical use cases for CDN include release of a product or update, and (near-)real-time coverage of an event. Downloading from carefully-maintained servers near the ultimate end-user can be many times as fast as comparable downloads from overtaxed, distant servers. CDNs play a vital, if specialized, role, and annual global revenues probably top US $2 billion.

Some commentators and businesses further restrict CDN to exclude peer-to-peer (P2P) technologies, or to consider “content” only in the narrow sense of audio or video files, but not HTML, JavaScript, or other Web objects. For our purposes, CDN emphasizes static content, and multiple locations.

ADN takes responsibility for dynamic delivery, including media streaming. Where CDN’s end-user populations often grow and collapse explosively, ADN focuses more on slowly-changing network topologies. CDN commonly models all endpoints–application servers and consuming end-users–within a single wide-area network (WAN). CDN leverages such specialized elements as WAN optimization controllers (WOCs) and application delivery controllers (ADCs).

CDN and ADN have much in common, including benefits of performance, demand pricing, security, and availability. They share many techniques, including TCP optimization, load-balancing, and caching. The boundaries between them are foggy, at best. Still, it’s important to understand the basics of each, if only to clarify your own organization’s requirements. Just as I’ve argued before that success with APM depends crucially on clear understanding of your own requirements and priorities, choice of a CDN or ADN provider demands careful analysis.

The relation between CDN and ADN, on one side, and APM, on the other, involves much more than analytic similarity. To measure your purchases of CDN or ADN, for instance, certainly demands a form of APM.

It will take several follow-ups to trace all the CDN-ADN-APM connections. In the meantime, consider whether your own deliveries are more like broadcast of static content, or interaction in a customized, user-specific way. We’ll see later how these two models carry different profiles for choice of APM.