Since the publication of my article “Evaluating Application Architecture, Quantitatively” in the 23rd issue of Microsoft’s The Architecture Journal , Iam receiving lots of questions / encouraging comments / wishes / suggestions. I never expected such a response back from the architects’ community around the world and result is this follow-up.
In the article ‘Evaluating Application Architecture, Quantitatively’ which is outlining the framework for evaluating application architectures quantitatively, it is been specified that for a positive response to every question / statement in the questionnaire / checklist '1' will be assigned and '0' will be assigned for a negative response. When a set of questions / checklist is used for an application architecture evaluation, some of them may not be suitable for a particular context.
Say for example, you are evaluating an application’s architecture that is meant for intranet only. So, in that context, assume that you are doing an architecture evaluation based on a particular repository of questions and it has a question which goes like this:
“Are web servers are placed in the DMZ zone?”
In this given context, this question is not applicable. For an intranet application, it is not a must to place the web servers in a DMZ zone. So, here if the response is “No” then zero is to be assigned against that question. But here the question itself is invalid or “Not applicable (NA)”. If the repository has more such “NA” questions, then resulting “Architecture Index” will be misguiding.
Although more no. of a questions make your repository rich and increase the chance of doing architecture evaluation for wider variety of applications, because of the nature of some of the contexts, some questions may become “invalid” or “Not Applicable”.
So, when you are building a tool, you should always have a provision to allow the reviewing architect to make a question as “Not Applicable” so that particular question will be excluded from the Architecture Index calculation.