Time-to-Release – the missing System Quality Attribute

re: Time-to-Release – the missing System Quality Attribute

Nick Malik — Sat, 07 Apr 2012 03:06:48 GMT

@Gerry,

Thank you for your pointer to the SABSA profile. Very thorough. In many ways, it is more complete than the profile available in the current ISO standards.

That said, it is as intellectually pure as the ISO list. The SABSA profile describes the product (as it was intended to) from different perspectives.

My post suggests that the project team needs to include a single "non-quality-attribute" in the list when performing the prioritization with the business customer. That single attribute, which I call "time to release" is a kind of lithmus test. Stuff "above" it on the priority list is sacred. Stuff below it is important, but not sacred. That defines the line between "we will not ship crap" and "we will ship imperfect software."

There is no perfect software.

re: Time-to-Release – the missing System Quality Attribute

Nick Malik — Sat, 07 Apr 2012 02:54:56 GMT

First off, I want to say that I'm honored to get a comment from someone as prominent and influential as Karl Weigers. Thank you, sir, for taking the time to read my humble blog.

I expected to get some criticism for suggesting that we include "time to release" on a list of quality attributes. You are correct in saying that it is not an attribute of quality. It is an indication of the amount of time and resources that a business is willing to commit to developing that product.

My intent was not to consider "time to release" as a quality attribute, per se, but to place this "non-attribute" on the same list as the other quality attributes when performing the prioritization with the business.

The goal, as I hope I explained, is simple: to empower the developers by making it clear, to EVERYONE, that some quality attributes are important enough to slip the ship date, while others are not. I cannot count the number of times I would have found that information useful.

In short, with respect to the ATAM method, I would say that the list of attributes should not be limited to descriptions of quality. Adding this one element would make a very big difference in the ability of the developers to prioritize their work and focus on the "stuff that matters the most" while, at the same time, resisting the often irrational pressure to "ship crap on time."

re: Time-to-Release – the missing System Quality Attribute

Gerry Giese — Fri, 06 Apr 2012 04:12:09 GMT

The SABSA methodology has the concept of "Business Attributes Profile" and has a taxonomy of business attributes. It's fairly well-rounded and extensive, even though it started from a security perspective. I have a copy of the book "Enterprise Security Architecture: A Business-Driven Approach" that discusses it, but I suspect you can find some or all of it online.

Good article. I too agree that time-to-release is important. Balancing tradeoffs is much more than trading off on technical attributes. If internal and external business -related attributes and context are not considered then the value proposition of a proposed architecture or solution goes out the window.

On a related note, I remember taking an SEI architecture course and the instructor saying that while you can propose quality attributes to a customer/stakeholder, always accept any quality attribute they dream up as long as they can tell you how they would measure it. The challenge then is to understand the push pull of the attribute against other attributes. Eliciting requirements, quality attributes included, is always a tough job, because you will never get all of them identified and prioritized, and even if you do some stakeholders will say they don't care about some of them. But as the saying goes, they don't care until they do. Process (such as architecture reviews) plus experience will take you furthest in the right direction.

re: Time-to-Release – the missing System Quality Attribute

Karl Wiegers — Sat, 24 Mar 2012 03:03:12 GMT

You point out that "time to release" is not found on typical lists of software quality attributes. That is because this is not a characteristic of *product* quality, but rather a characteristic of the *project* or of business objectives. Traditional attributes such as usability, maintainability, efficiency, availability, and the others are focused on properties of the product itself. I agree that "time to release" is an important success factor for a project, but it should not be lumped in with quality attributes that deliberately describe characteristics of the product being developed. Instead, I would use "time to release" expectations (based on some sound business analysis) to make appropriate trade-offs with features, quality, staff available, and budget available. It doesn't do you much business good to release a perfect product well after its effective market window, but nor does it usually do you much business good to release a piece of crap as quickly as you can.