- Needed to scale the website’s order processing capabilities to be able to handle 28k transactions in a 2-4 week period
- Existing integration infrastructure could not handle projected goals
- Hard deadline of 10 weeks for development and testing
- Take API-led connectivity approach: Create an API to expose customer and order information to SAP in real-time
- Design APIs using RAML and leverage OOTB components for faster delivery
- Use CloudHub to host APIs and provide secure connectivity to remote systems
- The solution was delivered on time and within project budgets and was recognized as a great success
- No issues were encountered during the peak season and the solution even exceeded processing expectations
- Seamless transition to the new business model created new channels of revenue and opportunities for the retailer
Shifting from B2B to B2C
This leading retailer spotted a significant opportunity to increase revenue by not only selling through a business-to-business (B2B) model for businesses ordering uniforms for their staff, but also directly to customers’ staff who wanted to order their own uniforms with a business-to-consumer (B2C) model. This meant that they had to shift from processing relatively few orders in high quantities to processing a very large number of individual orders.
Due to this change, they were anticipating significant increase in order volumes of up to 28,000 transactions for the initial 2-4 weeks from launch, requiring a transformation of their order processing capabilities. In addition, they had a hard go-live date of 10 weeks to ensure that the website was ready the step change in capacity.. This allotted them 6 weeks for solution development and environment provisioning and 4 weeks for user acceptance testing (UAT) and load testing.
While there was an established integration solution in place, it was not an ideal fit for the increase in volume and the cloud-focused nature of this project. To meet the challenge, the solution team proposed an API-led approach whereby they replace the on-premises SOAP web services and batch processing with a scalable website connected through consumer fulfillment APIs.
A Hybrid Platform Built to Scale
MuleSoft’s Anypoint Platform was selected to replace the existing solution because of its ease-of-use, rich API capabilities, native line-of-business connectivity and hybrid hosting capabilities.
The first step was to create a consumer fulfillment API designed to expose a series of APIs for registering customers and placing orders with an on-premises SAP instance. This API was consumed by the newly designed website hosted on AWS, which was built with scalability in mind to support the new B2C ordering model.
The APIs were designed using RAML, which was a key factor in ensuring that the website teams could quickly understand the API development requirements. To manage the load, they leveraged queuing and events from SAP’s master data catalogue. They also leveraged a number of out-of-the-box components and features including the SAP connector, REST API routing, JSON schemas, in-built reliability patterns, logging and notifications for a fast and consistent delivery.
Critical to the success of the project was Cloudhub, which hosted the APIs and provided secure connectivity to remote systems. CloudHub increased API reliability and significantly reduced development time and costs by providing built-in features such as zero downtime deployments, load balancing, persistent queuing, scalable workers, logging infrastructure and a cloud environment for provisioning which meant no operating systems nor dependent solution components had to be installed.
New Opportunities Previously Unavailable
Not only was the project completed within the aggressive 6-week development timeline, but an entirely new integration platform was established and the UAT and load testing signed off, all under 10 weeks. The speed to delivery and the increase in productivity were attributed to the intuitive and powerful tooling and platform provided by MuleSoft.
When they went live, no issues were encountered even during the peak times and the solution even went on to exceed expectations. During load testing, it was estimated that at a 300% peak, the API should be able to sustain 1,400 users per hour or 23 concurrent users. They later found out that during a peak time, the platform successfully processed 7,568 customer registrations and orders for 25 concurrent users in an hour.
During the first 5 months of operation, the API solution robustly handled a number of incidents such as network outages, and the platform gracefully handled aggressive AWS vulnerability patch schedules. Also, in more than one instance, exception logging features and CloudHub notifications detected problems far before other systems even noticed the issues.
For this leading retailor, Anypoint Platform enabled a reliable and seamless transition to the B2C business model that is continuing to establish new opportunities previously unavailable.