Our client is famous website to find personalized vihecle numberplate in UK. It has a database of 49,000,000+ personalized available numberplates. This website is using Solr for searching the matching plates. The query uses Solr regular expressions to find the all possible combinations of number plates for the user based on their provided keyword.
Client’s Solr implementation was having issue due to large amount of data and search loads. The Solr was crashing frequently due to memory issues. Whenever Solr crashed the search would come to a halt and caused lose in revenue. We did a performance audit to find the issues with Solr implementation. We found multiple issues with implementation which were main cause for the issues.
We did fine tune the schema for the cores and Solr configurations. We also upgraded the Solr to latest 6.x version. These steps did not only improve the performance but it also fixed the issues of Solr crashing. The final result of this project was increased availability of the search and increase in conversions. You can see the search in action here.