SQLite in 2025: Why This “Simple” DB Powers Major Apps

SQLite used to live in your browser or your phone. But in 2025, it's powering production apps — in the cloud. A new wave of tools is making it possible to use SQLite as a serverless, distributed, and production-ready database — without giving up its signature simplicity and speed. ... |**SELECT on 1 million rows**|~ **80 ms**|Typical SELECT query in a dataset with ~1 million rows. ([Toxigon][1])| |**INSERT operations per second**|~ **5,000 inserts/sec**|Under light/moderate load, single‐threaded or small concurrency. ([Toxigon][1])| |**UPDATE operations per second**|~ **3,000 updates/sec**|On the same 1M‐row dataset under similar test conditions. ([Toxigon][1])| … |**Small queries throughput: FTS5 vs LIKE vs others**|• FTS5: ~2,500–3,300 QPS, • LIKE operator: ~10–15 QPS, • “Normalized Table” / “Roaring Bitmaps”: much higher, tens of thousands QPS depending on filter complexity. ([Oldmoe's blog][3])|Tests on text search / tag filtering, showing the massive speed advantages when using full-text or optimized index/search methods vs naive LIKE. ... |**Many small HTTP requests / querying by ID**|• ~50,000 req/sec with pool size ~4–8 on certain implementations, • ~20-30k with lower pool/connections • Performance drops off with large pool sizes beyond CPU core count. ([GitHub][5])|These are web-server benchmarks running select by primary key, showing SQLite can handle surprisingly high request rates when optimized and with limited concurrency. ([GitHub][5])| |**Operation latency in GoatDB vs SQLite (simple operations)**|• Open DB (100k items): ~186 µs average, • Read item by ID: ~67 µs average, • Create table: ~1.1 ms, etc. ([GoatDB][6])| | … ## Can SQLite Be Used in Production? ... Let me address the elephant in the room. Every developer has heard the whispered warnings: "SQLite isn't for production," they say. "It can't scale," they insist. Well, I'm here to tell you that in 2025, those assumptions are not just outdated—they're costing developers opportunities to build simpler, faster applications. … ## Can SQLite Scale? The Truth About Performance Limits This is where the rubber meets the road. Scalability isn't just about handling more data—it's about handling more concurrent operations, more complex queries, and more demanding workloads without falling over. … - Unlimited concurrent readers - 100k SELECTs per second on properly tuned systems - In-memory caching for hot data **Write Performance:** Limited but Manageable - Single writer at a time (serialized writes) - SQLite can handle millions of QPS, and terabytes of data. However, our efforts to scale our Managed Jobs feature ran up against the one downfall of SQLite: many concurrent writers. … ### When Concurrency Becomes a Problem You'll hit SQLite's concurrency limits when: - More than 1,000 writes per second consistently - Write operations take longer than 100ms regularly - You have more than 50 concurrent writers - Real-time write consistency across multiple processes is critical … ### When SQLite Doesn't Win **Scale Requirements:** Thousands of concurrent writers will overwhelm SQLite's single-writer model **Distributed Systems:** Multiple servers need shared, consistent data access **Enterprise Features:** Advanced security, auditing, and compliance features may be limited **Team Collaboration:** Multiple developers working on shared data simultaneously