OMS implementation: increasing speed, consistency, and quality

• Client profile: European retail company 
• Project: develop a new order management system functionality and implement 

As part of the project, we were required to respond to business, operational and legal particularities with a customisable OMS. These can differ slightly or even totally in areas like FSP integration, providing customers with additional fiscal documents or different payment methods. 

• Extend functionalities and prepare roll-out at the same time: balance functionalities and development peaks to further extend the OMS and to maintain a “one platform approach”.  
• Ensure a smooth rollout execution: countries have different infrastructures, so executing the rollout requires complex tasks like infrastructure setup, initial data load or data migration.  
• Reduce release time by becoming a release independent solution: make an OMS able to adapt to legislation changes, business requirements. Getting fixes into production quickly and without dependencies was also among the challenges of the project.  
• Handle legacy test environments that generate inaccurate test data and results and didn’t allow for independent checks and finally releases.  
• Enhance the level of the quality for the new OMS, while reducing the total testing time.  
• Reduce the dependency of waiting for other solutions to perform integration tests with the OMS.  

Technologies

• Java EE 
• IBM Sterling Commerce 
• JBoss by Red Hat 
• Oracle Database 
• TIBCO 
• Angular 
• Dojo 
• Vagrant 
• Docker 
• Gradle 
• SonarQube 
• Jenkins 
• Groovy 

We tested the application on a standalone and integrative level, an approach that allowed us to start testing early from both a functional and performance point of view. During our early performance testing, we timely identified load problems which allowed the development team to fix them in a timely manner for a successful Black Friday.  

This enabled the platform to move from 140 orders/sec to a top of 400 orders/sec without any problems:

• Testing on multiple of the new OMS instances, on need and on scheduled time frames reduces time, effort and errors.
• Regression testing was reduced from 2 weeks to 3 days currently.

The aggregated order capability collects orders from multiple channels (online, call center and store)and then provides a single source of order status information across these channels.

This led to an increase in sales during Black Friday that had an average of 194% in orders from one year to the next. After the implementation, the system can have monthly independent releases for multiple countries. We have decreased the total time necessary for hotfixes from months to under 2 days, allowing enhanced system stability. 

The previous OMS solution generated a cost per transaction. While costs do exist with the new platform, these are not related to the number of orders. Leading to a decrease in costs, this was also great in the context of increasing the initial orders to a new 32K orders per day with a peak of 1 mill.