A social media platform improves search speed by an order of magnitude.
The speed of search can play a major role in either retaining users or turning them away to competitors.
Slow results or suggestions disengage the user quickly and force him to try out competitive websites. This is even more pronounced when there are plenty of similar websites available.
The client already had a working solution built on top of Solr (lucene based stack). They had tried many Solr optimizations but had not yielded the desired results. They were losing customers due to slow user experience.
The client engaged ThatNeedle to get improve the speed of search suggestion in the search box.
We at ThatNeedle discussed and analyzed the current architecture and problems faced by the platform. We suggested that trying to further optimize Solr would not yield the best results.
We recommended a decoupled microservice architecture that would run independent of Solr and the existing solution.
We created a fresh implementation of autosuggestions in python with a lean microservice architecture.
The new solution was 10 times faster than the previous solution based on Solr.
Solution highlights:
Based on our 10x faster implemention of the suggestions microservice, the user experience significantly improved and the client was able to achieve :