A diagram for cloud analysis: the big picture

The diagram in Jim Ditmore’s “Hard Truths About Cloud Differences” (HTACD) is a good starting point for a number of important points about cloud computing and especially its relation to application performance management (APM).

Don’t interpret it rigorously. The colors are purely aesthetic, and I have no idea why “Public Cloud” doesn’t extend to “Data Base”; there certainly is a thriving market in hosted database management systems. If you can relax about where individual trees are, and instead focus on the shape of the larger forest, HTACD gives a useful way to organize the complexity of cloud.

“Complexity” is one of the great ironies of this domain, of course. A central promise of cloud advocacy, which the diagram reinforces, is that cloud computing reduces complexity for organizations. A more mature perspective is that cloud is a tool for management of complexity, more than its elimination. As cloud usage normalizes, it’s inevitable that all the bubbles of the HTACD diagram will stretch to the right. We’ll come to regard this as an advantage, not a liability: while we consumers want software-as-a-service (SaaS) to appear on the left side of the graph for a satisfying out-of-the-box experience, we won’t stay there. To the extent that organizations rely on SaaS, for instance, for mission-critical operations, they’ll integrate the SaaS with other automations. Not only will SaaS be required to be simple, but it must also scale up to capture all the complexity of the organization’s real processes.

Another feature of the diagram is the population of the vertical axis. We sometimes have difficulty presenting the APM challenge to decision-makers–outsiders don’t grasp how much APM has to do to be successful. In arraying the labels “Application”, “Middleware”, …, “Network”, as it does, this diagram implicitly relates them to real business requirements. Consideration of this whole span helps open eyes to APM’s importance in reactively and predictively tracing upsets back to their root causes. One of “Application Monitor“‘s recurring themes through 2013 will be how tools need to be able to isolate problems wherever they originate: hardware faults, database misconfigurations, network transients, and so on.

As I mentioned earlier this week, part of APM’s development in 2013 will be reformation of licensing models to better match current realities of virtualization and cloud computing. Some APM products assume close control of “fat hosts”, and only fit the bubbles labeled “Co-lo” and perhaps “Private Cloud”. It’s inevitable, though, that more and more application value will arrive by way of SaaS and “Public Cloud”, and successful IT (information technology) departments will learn how to apply APM across these novel bases. We’ll return next week and in the months to come to study exactly those possibilities. In the meantime, what part of HTACD’s diagram is hot for you?