Video recording and production done by DevOpsDays.
PAYBACK GmbH operates Germany's biggest bonus programme PAYBACK, which has continued to attract increasing numbers of members and partner companies since it was started in the year 2000. The tremendous success story is continuing not only in Germany, but also in Poland, India and Mexico. Currently we are planning the rollout the PAYBACK Program in different other countries, e.g. Italy, France, Brazil and USA. PAYBACK sends out 85 million direct mailings, 350 million e-mail newsletters, 5 billion coupons distributed (1,4 printed, 3,6 digital) a year in Germany alone and its brand is represented in 36.000 partner stores. As a result, the programme is the largest multichannel-communications platform in Germany.
PAYBACK was acquired by American Express two years ago, therefore we are have ongoing internal challenges like the merger with AMEX. Furthermore, there are recent developments like rapid internationalisation, new partners, new channels like social media and mobile apps. We experience a growing complexity and size of our IT landscape with the demand for shorter release cycles and for new features while maintaining a high quality, performance and stability at the same time. Inspired by the DevOps movement and our first DevOpsDays 2011 in Goteborg we introduced a number of new ideas and tools which helped us to improve the delivery process and narrow the gap between Dev, Ops and QA. But during this transformation we realized that some of the ideas and tools that are spread by the DevOps community were either too simplified for us or we were not able to use them because of habits or constraints we have build up in the last 12 years.
In this talk we would like to share the experiences we've gathered during the last 2 years regarding our DevOps efforts. We are right in the middle of a transformation which we think is vital to our company. It turned out to be much harder and more time consuming than we thought at the beginning. We will talk about the issues with different data center providers with different processes and non trivial deployments for 14 applications across three-tiers in an Oracle-centric landscape and multiple servers plus configuration and database changes, challenges with ten globally distributed development teams, and of course, a cultural gap especially between Dev and Ops: different mentalities, responsibilities and levels of know how.
What we have learnt? A lot! And we would like to share it with you.
Most of the patterns and tools commonly suggested for adopting DevOps in your company are based on the assumption that you basically have a single platform/website. What do you do when you have the polar opposite?
At Future, we develop and run over a hundred different sites, ranging from small magazine support sites in WordPress, all the way through to high traffic, bespoke sites like TechRadar.com. They’re all written in different languages and platforms and have very different groups of developers with very different approaches and ideas, who’re based in several cities and countries.
This talk will detail how we introduced DevOps practices at Future: with the starting point of a chasm between Development and Operations, and entirely manual processes, we were able to introduce processes, team structures and tools that were flexible enough to fit the needs of our many different platforms and teams; increasing efficiency and reliability whilst keeping up with the commercial demands of the sites, which required more than 250 deploys a month before any automation.
The key to the successful transition was a ‘bottom-up’ approach, making incremental improvements to individual sites and changing how teams worked together and communicated, demonstrating the value to management at each stage and ensuring that commercial work was never delayed. This lead to a toolbox of useful systems and services that developers and operations engineers were able to use together to build appropriate workflows and processes for each different platform, but more importantly, a set of successful models for how to sell the values of this approach to each site.
After seeing how successfully this approach has worked for our websites, we have begun to introduce it into our traditional enterprise IT development. The mistakes, pitfalls and successes we have encountered along the way will be covered in detail, with a particular emphasis on in finding ways to get management and non-technical staff buy-in across the entire company.