Prior to GSPANN, the client’s other technology partner was managing the Solr search platform and was unable to solve the issue. After carefully analyzing the system architecture, our team understood that the complexity lies in extracting the inventory data (externalization), which was creating performance issues since the application was dealing with at least 1 million SKUs.
Our team proposed to perform externalization at an internal Lucene ID level by integrating Redis with Solr. We extracted the inventory information on a separate Redis environment and made Redis cache to save this information instead of Solr. This saved a lot of time consumed in replicating the database change on 200 Solr-based servers.
All inventory changes now flow to a single server where Redis is present and all servers are connected to the same Redis cache for inventory information.