Manticore Search for an universal search engine

Manticore Search, an open source high performance search engine project, joined The IT Press Tour this week to introduce their approach to this common challenge for a few decades now. Sergey Nikolaev, CEO and co-founder, took time to introduce the idea and gave us lots of details.

Everything started with the Sphinx project early 2000 but it stopped in 2016-17, so a new direction was clearly needed to leverage all the work done but also consider new challenges and needs with better performance, new technologies underneath, still in open source.

The firm was founded by Sergey Nikolaev, Peter Zaitsev, former CEO of Percona, and Mindaugas Zukas, COO of Altinity, leading a team of 10+ key developers.

The mission is to deliver a very simple scalable search engine, in open source mode, operating on affordable standard hardware. The team targets general search and log analytics with a real motivation to boost queries speed, resource consumption and SQL and JSON support. Beyond full-text search, it also adds faceted, Boolean, Fuzzy, Geo and Vector search, so very comprehensive model.

The demos we saw have been pretty impressive and fully transparent. The product is really very confidential but secured pretty names like Craigslist, Rozetka, Socialgist, Statista, Europrcs, Hotelplan, PubChem or Huispedia.

Several other products exist on the market like Elasticsearch, the table below explains a bit some differences:

They plan to add auto-sharding, authentication, integration with Kibana and auto-embeddings for vector search and consider some enhancements with AI.

The full source code is available on GitHub via this link.