Automatic provisioning and DevOps for NS

The NS Control 3.0 programme involves the replacement of the legacy system VKL by future-proof applications for adjusting personnel, timetables and equipment.

The challenge in this project

Control 3.0 involves an extensive and complex environment that requires a ditto complex and flexible development street (OTAP). Creating the different environments in the development street was very time-consuming. Also, setting up and keeping the configuration of the environments in sync was a precise and error-prone process. Especially when transferring the development teams to the operation, it is important to have a correct and representative environment available. The development and test environments are managed by the NS and the acceptance and production environments are purchased from Conclusion as a managed service environment.

In this context, we worked together on automatic provisioning on the one hand and OTAP management on the other. The aim of this process was to obtain one version of the truth about platform and configuration. And thus accelerate the speed of rollout of (changes) on environments and reduce the number of errors and/or disruptions.

Toegepaste technieken: GitHub, Jenkins, Puppet, Oracle Weblogic, Java, Oracle RAC, Maven, Ant, Linux.

How did Conclusion handle this?

The process started from a number of basic principles of Continuous Delivery. We used a model-based method to put together the platform. We used one central source to capture the model definition of the environment and ensured that both the Development Engineers and the Operations Engineers had a common goal for their environments.

This meant that the definition of the environment had to be entirely in a Puppet model, stored in a shared GIT repository. The Development Engineers and the Engineers responsible for the operation worked closely together in this process. By actually having them work together at the same location on the same platform and configuration, a common goal has been created. Also, the coordination on details of the solution was simple and fast.

By including the configuration of the current environment in one Puppet model, it is possible to build the environment at high frequency and test it for accuracy. In this way, the team worked iteratively towards a working platform and configuration. All changes are directly in the GIT repository, making everything traceable. Any change that leads to a non-working environment can be reverted within seconds.

As a result, the entire configuration is stored in a uniform manner and can be repeated for every step in the development line. The working method has also led to intensive cooperation between the developers of the application and the administrators of the platform. This team now works in a DevOps way.

Harmen Hummel, Control 3.0 program manager at NS: "My primary objective is to phase out our legacy VKL system as soon as possible. In 2016, automatic provisioning was recognised as a strategic area for improvement and Conclusion and operating company AMIS were commissioned to do so. Initially, this was a project-based approach, but this did not provide sufficient momentum for Control 3.0. In good joint consultation, we have therefore formed a joint Task Force with which we have achieved the desired results. We are now continuing this form of collaboration in the DevOps context. The bottom line is that we have gone from days to hours in terms of provisioning and deployment. All in all, I have a lot of confidence in this chosen path.

A great result

The result is a huge improvement in speed and flexibility in building and maintaining the various environments in the development street (OTAP). A complete environment consists of Oracle RAC, Oracle WebLogic, data feeds and multiple JEE applications as a highly available system across dozens of server nodes. Building a total environment can now be done within hours. In the past, this was several days. This speed gives a lot of flexibility in the use of environments. In addition, there is now full control over the configuration settings on each environment. Also, the promotion of changes by the OTAP street has become less complex, because the configuration is now fully automated.

This promotes communication and understanding of the environments between the developers and the operations, which has resulted in greatly improved collaboration (DevOps working) in addition to improving technology. This means that the Engineers can now proactively respond to changes, whereas in the past they had to search for the changes reactively. This has contributed to the quality, flexibility and job satisfaction.

Why did NS choose Conclusion?

Conclusion is able to offer a team that can support and automate the full set of products, network and underlying operating systems. This allows for quick action at the various levels and a specialist is always present to test and validate the chosen solution. Our team has a lot of DevOps knowledge to guide the transition and transfer the processes correctly. Also, our relationships with partners and suppliers (of software and supporting products) are essential to quickly resolve any disruption. This protects our customers from any pitfalls and unnecessary delays. We take responsibility for a working and effective end result.