Subtle but important: three smart performance pieces

Three of my favorite writers on performance topics recently made important points all having to do with finding the right balance in challenging circumstances. More than easy slogans quickly memorized, they bring deeper insight to technical topics.

Bob Wescott, for instance, author of Every Computer Performance Book, argues in “Working Close to the Edge” that “… the closer you are to some performance limit, the more careful you have to be with your calculations, models and predictions.” This particular article is a quick read–only 387 words. While the point might sound banal at first, it has important consequences for business planning. Organizations that leverage virtualization or cloud technologies, for example, to boost their efficiency, find they’ve necessarily taken on more risk. It’s not that virtualization is intrinsically risky at an engineering level; a predictable consequence in real-world organizations, though, is to highlight risks that were only latent before. Confusion about these matters leads to technology adoptions that lower efficiency and amplify risk.

“Portfolio management”, like virtualization, is a technique with a lot of business potential. I’ve never written about it myself because I’m wary about how easily it can go wrong. Christine Parizo explains well in “What application portfolio assessment and dog ownership have in common” how to achieve success with portfolio management. She aptly makes the point that, more than just buying the cute puppy of a rationalized application portfolio, the organization needs to make life-time adjustments to foster the healthy dog of ongoing portfolio balance.

Back at the front line of application performance, Tammy Everts asks “Are you making any of these 8 common (and wrong) assumptions about mobile performance?” Everts, “solution evangelist” for Radware, hits the nail squarely on the head, as always: mobile performance of the applications for which you’re responsible is likely both more important, and worse, than you realize. While this particular article doesn’t dive into the technical details of network protocols I assume she knows, she reinforces that mobile performance needs its own measurements. Non-linear interactions can, and often do, degrade performance unexpectedly. True end-to-end application performance management (APM) measurement is necessary to capture many of these effects. Otherwise, it’s easy to have servers and networks that appear to be doing their job in delivering responsive applications, but actually exacerbate the frailties of wireless networking to frustrate end users. Don’t let this happen to you.