Continuously Delivering greenfield and legacy applications en masse
I recently gave a talk at Agile Horizons 2012 on behalf of my amazing employer Sky Network Services, detailing our yearlong Continuous Delivery journey and the evolution of our Strangler Pipeline. As a follow-up I intend to write a series of articles on our pipeline, as it is a narrative far removed from the “pipelining a single greenfield application” model often found in Continuous Delivery experience reports.
Sky Network Services is an agile, innovative technology company that produces telecommunications middleware for BSkyB. Despite a plethora of talented technical/non-technical staff and an enviable reputation for delivering quality software, an in-house analysis in mid-2011 identified a number of problems:
- Many applications used different methods of deployment, start, stop, and database migration in different environments
- There was little visibility of which application versions were progressing through the test environments at any given time
- Releasing a minor bug fix for an application necessitated a re-release of the parent platform
- Development teams and Operations teams were constrained to separate silos
At this point we were attracted to the Continuous Delivery value proposition, albeit with the additional challenge of scaling our pipeline to manage an estate of legacy/greenfield applications that in the past year has doubled in size.
In this series of articles I aim to cover:
- Challenges – how we solved the more common Continuous Delivery challenges
- Scaling Up – how we scaled our pipeline to manage our ever-growing application estate
- Legacy and Greenfield – how we simultaneously release legacy and greenfield applications
- Autonomation – how we established a Continuous Delivery transformation across a Waterscrumfall organisation