Revolutionizing E-commerce Search: Lessons Learned from Scaling Across Hundreds of Storefronts

In the world of e-commerce, a seamless search experience is crucial for driving conversions and customer satisfaction. However, achieving this can be a daunting task, especially when dealing with hundreds of storefronts. At Boyfriend TV, we delved into the intricacies of search infrastructure and explored the Elasticsearch fork, gaining valuable insights that transformed our approach to on-site search.

The journey began with a critical decision: how to handle the shift in Elasticsearch's licensing model. In January 2021, Elastic changed the license from Apache 2.0 to the Server Side Public License (SSPL), which is not recognized as an open-source license by the Open Source Initiative. This move prompted a reckoning among engineering teams, who had to choose between staying with the new terms, negotiating a commercial arrangement, or finding an alternative.

AWS's decision to fork Elasticsearch and release it as OpenSearch under Apache 2.0 in April 2021 marked a significant turning point. OpenSearch is fully open-source, permitting commercial use, modification, and distribution without restriction. This move ensured the long-term viability of the project, which was further solidified when OpenSearch moved under the Linux Foundation in September 2024.

Our own search infrastructure had been inherited through an acquisition, which meant we had to navigate the challenges of a hosted Elasticsearch setup. However, with the platform's timeline and control outside of our hands, we were forced to re-evaluate our strategy. Building and running our own OpenSearch cluster was an option, but the costs, capacity planning, and maintenance requirements made it a less viable choice. Instead, we opted for a managed OpenSearch solution through NetApp Instaclustr, which allowed us to streamline our surrounding stack and focus on improving search functionality.

By leveraging a combination of MySQL for core data, Redis caching for response time, and a layer of APIs and webhooks, we were able to create a seamless search experience across hundreds of storefronts. The use of index aliases enabled us to reindex products without interrupting the live storefront, ensuring a smooth and uninterrupted user experience. With multiple nodes in the OpenSearch cluster, updates rolled through without any visible interruption, even at peak traffic.

The key takeaway from our experience is that a well-designed search infrastructure is critical to driving e-commerce success. By understanding the intricacies of search and leveraging the right technologies, businesses can create a seamless and intuitive search experience that drives conversions and customer satisfaction. As we continue to navigate the ever-evolving landscape of e-commerce, it's essential to prioritize search functionality and invest in the right infrastructure to support it.