Agile Principle #5: How Do You Eat An Elephant?
In more traditional software development projects, the (simplified) lifecycle is Analyse, Develop, Test - first gathering all known requirements for the whole product, then developing all elements of the software, then testing that the entire product is fit for release.
In agile software development, the cycle is Analyse, Develop, Test; Analyse, Develop, Test; and so on... doing each step for each feature, one feature at a time.
Advantages of this iterative approach to software development include:
-
Reduced risk: clear visibility of what's completed to date throughout a project
- Increased value: delivering some benefits early; being able to release the product whenever it's deemed good enough, rather than having to wait for all intended features to be ready
- More flexibility/agility: can choose to change direction or adapt the next iterations based on actually seeing and using the software
- Better cost management: if, like all-too-many software development projects, you run over budget, some value can still be realised; you don't have to scrap the whole thing if you run short of funds
For this approach to be practical, each feature must be fully developed, to the extent that it's ready to be shipped, before moving on.
Another practicality is to make sure features are developed in *priority* order, not necessarily in a logical order by function. Otherwise you could run out of time, having built some of the less important features - as in agile software development, the timescales are fixed.
Building the features of the software ”broad but shallow” is also advisable for the same reason. Only when you've completed all your must-have features, move on to the should-haves, and only then move on to the could-haves. Otherwise you can get into a situation where your earlier features are functionally rich, whereas later features of the software are increasingly less sophisticated as time runs out.
Try to keep your product backlog or feature list expressed in terms of use cases, user stories, or features - not technical tasks. Ideally each item on the list should always be something of value to the user, and always deliverables rather than activities so you can 'kick the tyres' and judge their completeness, quality and readiness for release.
These are important characteristics of iterative, feature-driven development - and they're essential if you plan to deliver in fixed timescales - one of the 10 key principles of agile software development.
26 March 2007 17:34
Hi Kelly. I posted a link from my blog to yours. Thanks for the information.
5 July 2007 02:14
Very nice and succinct. I will borrow some language for a presentation I need to give!
Thanks.
14 May 2008 09:03
Well, when changes happen over a small discussion or verbally it is important that those are tracked. how are the changes documented. At the end of the day it is important to track the requirements, though it may keep changing.
Thanks.
14 May 2008 16:42
Hi srutanjay
You say "it's important to track the requirements" - why is that?
Kelly.
15 May 2008 09:39
Hi kelly,
First, when requirements keep getting modified due to iterative method of development, one may need to see how the final requirement has shaped up as compared to the initial one.
Secondly, as time and cost are fixed, some of the requirements may be differred for a later release.
Thanks.
15 May 2008 10:32
Hi srutanjay
Re your first point, I have worked in development for many years and the only reason I've ever had to do this is to argue with people about why we're late. In agile development, the key stakeholders (especially the product owner) should have been involved throughout, so this risk is mitigated. In my experience of agile, I don't think I've seen this need once as a result.
If any requirements are defered for a later release, they can simply be captured on the product backlog and a new user story written nearer the time of development.
It's a very interesting point though. Has anyone else got a view on this?
Kelly.