Impact!

A Practical Approach to Large-Scale Agile Development

This is a strange one. A Practical Approach to Large Scale Agile Development: How HP Transformed LaserJet FutureSmart Firmware is a 170 page case study of an agile transformation at a company that's not typically associated with modern development practices, with 400 firmware developers - not exactly the typical use case for agile either. The case study is presented from a high level management perspective, and covers topics such as continuous integration at scale, overcoming cultural and organisational barriers to implement iterative planning and estimation, but also stuff rarely covered in books with the word "agile" in the title, for example cultural issues for cross-continent development between India and US.

The case study is very interesting, and I'd recommend reading through the relevant chapters to anyone trying to push agile processes to a large organisation, as it describes two aspects that I see very rarely, but think are incredibly important:

1. The process change was started with a clear business outcome in mind. Not because agile is cool, not because everyone else is doing Scrum, but because the company wanted to achieve a measurable, quantified reduction in development costs and reallocate funds to innovation. This story shows how having a clear goal and driving a process change with that can help to make the right decisions and adjust the process to a particular environment, at the same time aligning everyone in the organisation on why and how to change the process. The second chapter opens up with a great quote 'You should be agile not just to be agile, but to drive the business results' and advises that companies describe clearly what they want to achieve from an agile transformation before embarking on that journey.
2. The change was driven by principles, and practices were evaluated and widely modified to achieve the set business goals. Unlike many teams out there who slavishly follow practices but discard principles and end up with no benefits, these guys actually invented their own process driven by agile principles. As a result, they claim to have reduced the development costs by 70%.

Before really changing the process, they restructured their software to support iterative delivery, which is another great lesson. Without a way to ship things out and get feedback iteratively, in reasonably short cycles, very little other stuff from the agile toolbox works. Another interesting aspect of their transformation is how they applied process metrics, which is a common problem with large organisations. Their approach is to use metrics as a conversation starter: 'The key is not to manage by metrics, but use the metrics to understand where to have conversations about what tis not getting done'

A negative side of this book for me is a very liberal use of terminology. Although there is a lot of guilt-by-association with Scrum concepts, and the authors say it is 'quite like Scrum', my understanding from the book is that they ended up with a process that is their unique take on the topic and any links to Scrum were really pushing it. I don't mean to argue that their process was wrong - it obviously works for them and provides a lot of value, which is the only thing important - but many Scrum references just felt out of place. I felt that Scrum got equated with just running iterations, and there is a lot more to a good Scrum process. For my taste, there is just too much management speak, including 10x productivity increases, quantum leaps, and of course sprinkles of SCRUM capitalised as if it were an acronym. This is why I'll give the book only 4 stars.

A Practical Approach to Large-Scale Agile Development
Verdict:         
Authors: Gary Gruver, Mike Young, Pat Fulghum
Publisher: Addison-Wesley Professional
ISBN: 978-0321821720
Pick it up from Amazon UK or Amazon.Com.