Fluidra Makes Waves With Real-Time, Customer Advocacy

As market leaders, Fluidra knew that they needed to keep innovating whilst offering a scalable, performant, and reliable service. They needed to process data much faster to support an augmented personalized customer experience – they wanted to perform calculations and run algorithms against prices immediately while interacting with their customers in real-time. They also needed to integrate with other systems like Salesforce and logistics platforms. This prompted the design of a new architecture that would significantly accelerate data processing as well as decoupling from systems that were inflexible and costly to maintain. They started to look at in-memory solutions that could deliver data immediately to the new services that they were building. Several solutions were reviewed by an external consultancy, but Hazelcast was ultimately recommended. This prompted discussions with Hazelcast partners who assisted with building a new architecture that was decoupled from the ERP systems allowing them to build real-time, front-end services. The e-commerce platform was the first service deployed as this was now the only way that consumers could have access to products, as the pandemic forced the closure of all stores and shops. Core to the overall architecture is a centralized, decoupling layer that is isolated from the ERP systems. Within the decoupling layer, there are 3 key components: 1. The enterprise service bus – for locating data 2. Data staging – for high-speed data storage and retrieval 3. Event streaming – to communicate with internal services The Hazelcast in-memory data store is used in the data staging layer for high-speed data storage and retrieval. Other database platforms (like MongoDB) will also be hosted there, some of which currently do not require real-time performance. The data staging layer will provide all data access requests for all applications via APIs, whether it is for the data warehouses, the ERP systems, Web sites, and e-commerce. The Hazelcast stream processing component is used for the loading of stocks and prices for all products stored in the shopping baskets in real-time. The ERP MDM system pushes customer, product, and pricing information data to the event streaming system that runs on AWS Kinesis and Hazelcast then reads the data and qualifies it. Hazelcast was a crucial component in ensuring that as stock and prices changed, e-commerce customers had a real-time view of what was available and at what price as they made inquiries. All the pricing logic is hosted on Hazelcast to dynamically calculate unique pricing for each customer and product combination. As they expand their system, Fluidra is planning to integrate other data sources using Hazelcast stream processing. Having in-memory storage and stream processing capabilities within a single platform creates an architecture known as a “digital integration hub.” This architecture allowed Fluidra to integrate data from multiple sources and process data in real-time, without worrying about integrating separate solutions from separate solution vendors. This mitigated the complexity caused by integration, performance, and management issues, while also avoiding separate licensing costs.