Met Gloria Vinogradova from the Manticore Search team in Novosibirsk. Rest of the crew — next time. Good start though. Real talk for everyone who's followed me a while: yes, ClickHouse is my one love. I run it daily at tens of petabytes. But being in love with one tool doesn't mean going blind to the rest — and I'd know a thing or two about blind. Here's what the ClickHouse stans on this app don't want to hear: it is NOT a search engine. It's an analytics monster. Columnar storage, aggregations that make billions of rows beg for mercy. But the second your task is full-text search, fuzzy matching, relevance, ranking — ClickHouse is the wrong hammer and you're just bruising your own thumb. That's where Manticore walks in. Open-source, C , near-total rewrite of Sphinx. Full-text search, real-time indexing, vector search, columnar storage too. SQL MySQL protocol. Their benchmarks put it way ahead of Elasticsearch on log analytics — and it runs on 1 core and 1GB of RAM while Elastic is over there asking for another JVM heap dump and your firstborn. And no, I'm not talking out of a docs page. I ran one of the biggest Manticore clusters I've ever laid hands on: 40 nodes, 20 shards, pushing 1 petabyte, chewing tens of TB of logs a DAY. Rock-solid access, blackjack, the whole circus. You push that kind of volume through a search engine and it doesn't flinch — you respect it. The lesson nobody on tech twitter wants to admit: there is no one database to rule them all. ClickHouse for analytics. Manticore for search. Postgres for transactions. An engineer worth a damn doesn't simp for one technology — they know how to balance and how to actually cook each one. Picking the right tool is the job. Stretching your favorite DB over everything isn't engineering, it's cope. I love ClickHouse. But when the task is search, I grab what was built to search. Try having a personality beyond your favorite database. Shout-out to the Manticore crew — you're shipping something real. Keep the bar high. github.com/manticoresoftware… #ClickHouse #ManticoreSearch #BigData #DataEngineering #OpenSource #SearchEngines #Postgres #Elasticsearch #DevOps #Databases

🚀 New interactive course by Manticore Search! Master Morphology & Lemmatization: stemmers, lemmatizers, morphology settings and how to boost your full-text search quality. No setup needed — learn directly in your browser! 👉 play.manticoresearch.com/mor…

🎓 New interactive course: Escaping in Manticore Search! Master how to properly escape special characters and write bulletproof search queries. Real-time practice with instant validation ✅ 👉 play.manticoresearch.com/esc…

🚀 New Interactive Course: Auto-Embeddings & Hybrid Search with Manticore Search! Learn how to combine full-text and semantic (vector) search in one powerful query. Automatic embeddings, hybrid_match(), RRF fusion, real-time indexing and more — all hands-on! Start right now👇 play.manticoresearch.com/aut…

🔍 "More Like This" started as lexical search: you take a document 📄, extract important words and find similar texts. Today the same scenario has evolved into embeddings 🧠, ANN indexes and hybrid search. We explored this evolution in the article 👇 manticoresearch.com/blog/the…

🚀 What if your vector search could be significantly faster without sacrificing accuracy? In our latest deep dive, we show how adaptive early termination lets HNSW stop searching when further exploration is unlikely to improve results — cutting unnecessary work while preserving high recall. If you're building AI search, RAG, or recommendation systems, this is worth a read.👇 manticoresearch.com/blog/knn…

🚀 New interactive course by Manticore Search! Master table settings: plain vs real-time tables, rt_mem_limit, morphology, stopwords, html_strip, columnar storage and more. Just 15 minutes to make your tables faster and more relevant. 👉 play.manticoresearch.com/tab…

Do you remember when you joined X? I do! #MyXAnniversary

vector search misses what your users actually type. building Manticore Search we see this constantly: error codes, exact identifiers, version numbers, rare entity names. hybrid (BM25 vector RRF) recovers them without retraining your embedding model.

🔍 Now xt850 matches xt 850! In Manticore Search 23.0 , bigram_index lets users search model names without spaces — and still get correct matches. Perfect for product catalogs: xt850, galaxy24, iphone5se, eos80d, etc. Learn more in the new post 👇 manticoresearch.com/blog/how…

🚀 How to speed up phrase search in Manticore Search by 3x? New post: bigram_index delivers up to 2.9× higher QPS and 3.2× lower latency on a million documents. We break down all modes (all, first_freq, both_freq) and when to use each. Read more 👇 manticoresearch.com/blog/how…

🚀 New demo blog post: how to quickly build search that doesn’t feel like a placeholder. Autocomplete, typo handling, filters, similar items, semantic search — not as separate features, but as one working flow. A simple example with ready-to-use code 👇 manticoresearch.com/blog/man…

New interactive course: Secondary indexes in Manticore ⚡ It should answer all your questions about how it works 👉 play.manticoresearch.com/sec…

Regex in Manticore — now in an interactive course 🧠 Practice with real examples 👉 play.manticoresearch.com/reg…

Explore Manticore’s columnar storage in a new interactive course 📦 Learn how it works with hands-on practice 👉 play.manticoresearch.com/col…

New interactive course: joins and JSON in Manticore 🔗 Learn how to combine and query data effectively 👉 play.manticoresearch.com/joi…

Fuzzy search made simple 🔍 Try the new interactive Manticore course and learn by doing 👉 play.manticoresearch.com/fuz…

Search issues in production rarely start with a dramatic outage. More often, performance just slowly slips 📉 This new Manticore Grafana dashboard helps you catch that early, before it turns into a bigger incident. Read more → manticoresearch.com/blog/gra…