quality thriving 2017

Sphinx Manticore Search

When Sphinx Search development stalled in 2016 and bugs went unfixed, former Sphinx team members reunited to fork it as Manticore Search. Sphinx later went closed-source, vindicating the fork.

What it is

Manticore Search is a full-text search engine written in C++ that supports SQL-like syntax (SphinxQL), real-time and plain indexes, and integrates with MySQL protocol. It offers sub-second search over billions of rows and supports features like faceted search, geo-search, and morphology processing.

The story

Sphinx Search was once the go-to full-text search engine for anyone who found Elasticsearch too heavy and Solr too enterprise-y. Written in C++, it was blisteringly fast and beloved by the MySQL crowd for its SphinxQL interface. But by late 2016, development ground to a halt. Bugs piled up unfixed, promised features never materialized, and support customers started getting nervous.

In May 2017, a group of former Sphinx employees — who had scattered to various companies after the original team dispersed — decided to reassemble. They forked Sphinx 2.3.2, attracted investment, and launched Manticore Search. The team managed to gather back most of the previous Sphinx team members, which gave them both the institutional knowledge and the credibility to pull off the fork.

The timing proved prescient: in December 2018, Sphinx resumed development with version 3.0.1, but with a twist — it was no longer open source. The last open-source version of Sphinx remains 2.3.2, the exact version Manticore forked from. The original project's move to closed source was the ultimate validation of the fork's decision.

Manticore has since evolved far beyond the original Sphinx codebase, adding features like real-time indexes, columnar storage, secondary indexes, and a Kibana-like web UI called Manticore Buddy. It positions itself as a lightweight, fast alternative to Elasticsearch that's particularly well-suited for log analytics and e-commerce search. Three years after forking, Manticore was processing search differently enough that going back was never a consideration.

The Manticore story is the textbook case of 'the team is the project, not the name.' When the people who built the software leave and reconvene under a new banner, the old project becomes a shell.

Timeline

Sphinx Search initially released as an open-source full-text search engine

Sphinx development stalls; bugs go unfixed, support lapses

Former Sphinx team members fork Sphinx 2.3.2 to create Manticore Search

Sphinx resumes development as closed-source with version 3.0.1

Manticore introduces columnar storage engine and positions as Elasticsearch alternative

Key people

Sergey Nikolaev
Manticore Search co-founder, former Sphinx developer
Andrew Aksyonoff
Original creator of Sphinx Search engine
Manticore Steering Committee
Committee of experts and heavy users guiding the project's direction

Impact

Manticore Search preserved and extended one of the most efficient full-text search engines ever built, preventing it from disappearing into proprietary obscurity. For organizations that had built their search infrastructure on Sphinx, Manticore provided a seamless migration path with active development and support.

The fork also demonstrated that when the original developers are the ones doing the forking, the fork has a massively higher success rate. Manticore didn't just keep Sphinx alive — it evolved it into a competitive alternative to much larger projects like Elasticsearch.

Lesson: When the whole team forks and the original goes closed-source, there's no ambiguity about who carries the torch.

Related forks