Lori MacVittie, Senior Technical Marketing Manager at F5 Networks (www.f5.com), says:

Like most technology jargon, there are certain terms and phrases that end up mangled, conflated, and generally misapplied as they gain traction in the wider market. Cloud is merely the latest incarnation of this phenomenon, and there will be others in the future. Guaranteed.

Of late the term “application delivery” has been creeping up into the vernacular. That could be because cloud has pushed it to the fore, necessarily. Cloud purports to eliminate the “concern” of infrastructure and allows IT to focus on … you guessed it, the application. Which in turn means the delivery of applications is becoming more and more pervasive in the strategic vocabulary of the market.

But like cloud and its predecessors, the term application delivery is somewhat vague and without definition. I am not going to define it, in case you were wondering, because quite frankly I’ve watched its expansion and transformation over the past decade and understand that application delivery is not static. As new technology and deployment models arise, new techniques and architectures must also arise to meet the challenges that naturally arise along with those applications.

But how, then, do you know what is and is not application delivery? If it can morph and grow and transform with time and technology, then anything can be considered application delivery, right?
Not entirely. Application delivery, after all, is about an end-to-end process. It’s about a request that is sent to an application and subsequently fulfilled and returned to the originator of the request. Depending on the application this process may be simple or exceedingly complex, requiring authentication, logging, verification, interaction of multiple services and, one hopes, a wealth of security services ensuring that what is delivered is what was intended and desired, and is not carrying along something malicious.

A definition comprising these concepts would be either be far too broad so as to be meaningless, or so narrow that it left no room to adapt to future technologies. Neither is acceptable, in my opinion. A much better way to understand what is (and conversely what is not) application delivery is to learn three simple axioms that define the core concepts upon which application delivery is based.

“Applications are not servers, hypervisors, or operating systems.”
Applications are not servers. They are not the physical or virtual server upon which they are deployed and from where they draw core resources. They are not the web and application servers on which they rely for application-layer protocol support. They are not the network stack from which they derive their IP address or TCP connection characteristics. They are uniquely separate entities that must be managed individually.

The concrete example of this axiom in action is health-monitoring of applications. Too many times we see load balancing services configured with health-checking options that are focused on IP or TCP or HTTP parameters. Ping checks, TCP half-open checks, HTTP status checks. None of these options are relevant to whether or not the application is available and executing correctly. A ping check assures us the network is operating and the OS is responding. A TCP half-open check tells us network stack is operating properly. An HTTP status check tells us the web or application server is running and accepting requests. But none of these even touches on whether or not the application is executing and responding correctly.

Similarly, applications are not ports, and security services must be able to secure the application, not merely its operating environment. Applications are not – or should not – be defined by their network characteristics, and neither should they be secured based on these parameters.

Applications are not servers, hypervisors, or operating systems. They are individual entities that must be managed individually, from a performance, availability, and security perspective.

“Availability, performance, and security are not separate operational challenges.”
In most IT organizations the people responsible for security are not responsible for performance or availability, and vice-versa. While devops tries to bridge the gap between applications and operations-focused professionals, we may need to intervene first and unify operations. These three operational concerns are intertwined, they are interrelated, they are paternal triplets. A DDoS attack is security, but it has – or likely will have – a profound impact on both performance and availability. Availability has an impact on performance, both positive and negative. And too often performance concerns result in the avoidance of security that can ultimately return to bite availability in the derriere.

Application delivery recognizes that all three components of operational risk are inseparable, and they must be viewed as a holistic concern. Each challenge should be addressed with the others in mind, and with the understanding that changes in one will impact the others.

“Application delivery decisions cannot be made efficiently or effectively in a vacuum.”
Finally, application delivery recognizes that decisions regarding application performance, security, and availability cannot be made within a vacuum. What may improve performance for a mobile client accessing an application over the Internet may actually impair performance for a mobile client accessing the application over the internal data center network. What is appropriate authentication methods for a remote PC desktop are unlikely to be applicable to the same user requesting access over a smartphone. The various components of context provide the means by which the appropriate policies are enforced and applied at the right time to the right client for the right application.

It is context that provides the unique set of parameters that enfolds any given request. We cannot base decisions solely on user, because user may migrate during the day from one client device to another, and one location to another. We cannot base decisions solely on device, because network conditions and type may change as the user roams from home to the office and out to lunch, moving seamlessly between mobile carrier network and WiFi. We cannot base decisions solely on application, because the means and location of the client may change its behavior and impact delivery in a negative way.

When you put these axioms into action, the result is application delivery. A comprehensive, holistic and highly strategic approach to delivering applications. It is impossible to say application delivery is these five products delivered as a solution because whether or not those products actually comprise an application delivery network depends on whether or not they are able to deliver on the promise of these three axioms of application delivery.