A Natural Successor to Sphinx for Large-Scale Search

[il76]

I've been working with search technologies since the early days of Sphinx Search and eventually migrated all of my projects to Manticore Search. Today, I primarily use RT tables, some percolate functionality, and the MySQL protocol, indexing databases with around 30 million records for both full-text and attribute-based search.

What I appreciate most is the MySQL protocol itself. The familiar SQL syntax makes Manticore easy to work with, and I much prefer this approach over HTTP APIs. Performance is another strong point. While I found percolate somewhat slow for my particular workloads a few years ago, I haven't revisited it recently, so that may well have improved.

Resource efficiency is also a major advantage. We previously used Elasticsearch, but it required more hardware resources and was significantly more complex to administer. Manticore has proven much easier to operate while still delivering the search capabilities we need.

At the moment, I don't feel that I'm missing any important features. The platform covers my requirements well. The only challenge I remember was with replication. I had difficulty keeping it stable, especially when changing index schemas by adding or removing columns. That experience was some time ago, though, and things may have changed since then.

My overall experience has been very positive. When Sphinx development slowed down, I noticed Manticore appearing as a fork and decided to evaluate it. It quickly proved itself as a worthy successor, and over time I migrated everything to Manticore.

I first learned about the project through articles on Habr when Manticore was being introduced as the next generation of Sphinx.

Ilya Gurylev,
CTO in Lawyer Social Network 9111.ru

[airolg]

Thank you, Ilya, for your trust and for staying with us from the Sphinx days through your migration to Manticore. It's especially rewarding to hear that Manticore is powering search across tens of millions of records at 9111.ru while remaining efficient and easy to operate. We also appreciate your feedback on replication and percolate performance — real-world experience like yours helps us continue improving the product.