Fast and reliable search engine for massive datasets

[YurovY]

I use Manticore Search as the core engine for storing and searching through large datasets — around 37 million XML records from the Russian company registry (ЕГРЮЛ) and address data from FIAS, split across several tables. I mainly work through the SQL-like interface, using fuzzy search and key-based lookups.

What I like most:

1. Very high speed of fuzzy search, even on tens of millions of records.
2. Easy integration thanks to the familiar SQL syntax.
3. Straightforward setup and reliable performance under production loads.

What could be improved:

1. Better backward compatibility — updating from version 6.3.3 to 8.1.1 caused “connection reset by peer” errors that were hard to debug since no errors appeared in the server logs. It might have been memory-related (I increased RAM to 48 GB, but the issue still appeared occasionally).
2. More detailed diagnostic tools, like extended logs or network-level tracing.
3. Built-in performance monitoring utilities.

As for improvements, it would be great to have:

• Native query profiling tools to analyze complex queries.
• A clearer migration guide between major versions, including possible breaking changes.

Overall, my experience has been positive. For full-text search tasks, Manticore delivers fast performance, simple configuration, and minimal dependency on external services.

I discovered Manticore while researching tools for fast fuzzy search across large datasets.

Yaroslav Yurov
Data Kaizen Limited Liability Company

[airolg]

Thank you for such a detailed and thoughtful review!
It’s great to hear that Manticore performs well for your large-scale datasets like ЕГРЮЛ and FIAS, and that you appreciate the SQL interface, fuzzy search speed, and ease of setup.

We also really appreciate your valuable technical notes. Your feedback on version migration, diagnostics, and performance monitoring is very helpful — these are exactly the kinds of insights that help us improve Manticore for production-scale environments.

If you ever encounter similar issues again (like the “connection reset by peer” problem), feel free to share more details on our forum or GitHub — we’ll be glad to look into it with you.

Thanks again for choosing Manticore and for taking the time to share your experience. We’re happy to have users like you pushing Manticore to handle such serious workloads!