Broader industry support of APDEX needed!

Rather than just talking about the end users’ perspective, WildPackets is addressing it head-on by being the only network analysis software vendor to implement the APDEX approach, as suggested by Doug MClure in his comment. By leveraging the work of the APDEX initiative, WildPackets is able to identify each and every transaction that exceeds the threshold for end-user tolerance. These include real-world transactions, like how long it takes to download a web page or how long it takes to get a response from a corporate database. This is the measurement that matters to all of us as end-users, and APDEX allows us to derive, report and alert on what users are really experiencing.

The key word is "end user" -- what they experience the application to be is, ultimately, what the performance of the application.

What they consider good or what they consider frustrating is what it ultimately is.

The question becomes how do you measure and then do something actionable about it!

There are many kinds of users, the end-user, the company, another company using your services, even an organization and departments. Each has different requirements, different effect to systems and "user" satisfaction, etc. And then there is the most important "user" - the business (and profit) goals of you company!

So - not so simple! APDEX is one way to formalize the performance measurement but.. Asking an "application" performance is a little too wide question today - previously an application was just maybe "one" program running or a bunch of transactions. Today an application touches (most of the time) several parts of the whole infrastructure and that's where the problem starts.

Too many companies still live without (business) capacity planning or really don't understand what it means. They just run performance numbers, try to speed up one part of the whole and often end up even worst problems they had previously.

Yes, most important is the "user" but, as in computers, priority alone can not be used for performance management because nobody has unlimited resources. Yes, like largest companies, simulations and tests can be done, but how many know how? Technically the most important, I think, is not to corner yourself without plan B. If usage performance would be easily predicted we could do the same with stock and start trading futures.