How can you make your Continuous Delivery 50 times faster?

Written by Wilfried Kirschenmann, on 28 August 2019

This year, ANEO is renewing its participation in the DevOps REX! On this occasion, we are sharing our feedback on concrete DevOps cases.

We encountered the Digital Banking department of a major French bank that was suffering from a chronically unstable CI/CD pipeline associated with a decreasing yield.

The release management teams responsible for the industrialization of deliveries and deployment for the Digital Banking scope applications had implemented a semi-automated software factory to allow developers to deliver their applications to production with as little human intervention as possible.

Organizational barriers and poor technical and procedural choices were hindering the operation of the software factory. Added to this were time-consuming maintenance and malfunctions that eroded the development teams' trust in the existing setup.

We were then approached to conduct a rapid diagnosis to identify the weaknesses in the process and the implemented system, and to provide an alternative to stabilize the delivery chain and perfect an efficient and stable one-click process.

Our intervention was divided into three parts:

  • A flash diagnosis to understand the setup as a whole and highlight the reasons for the slowness of the chain.
  • The definition of the architecture of the alternative solution, specifying the process and tools to be implemented based on the client's context and our expertise in market solutions.
  • The creation of an operational prototype that will serve as a basis for the target implementation, and which allows in advance to validate the promises brought by the alternative solution and to identify additional work areas that would need to be addressed during the final implementation.

This project was marked by very high expectations, a delicate geographical context (teams spread across various countries in Europe and Asia), and a somewhat skeptical atmosphere tinged with the fear of seeing a new complicated setup emerge.

Yet, the prototype of this new chain won the approval of all teams from the first demonstration. And for good reason, those chosen as pilots to test the prototype immediately noticed better reliability of the system, almost zero human intervention, and a 50% reduction in application delivery times. QED!