Warning: Queries may be slow during periods of high write throughput or after a large bulk import.
turbopuffer can handle >= 10,000 writes/s (WPS) per namespace, but indexing cannot currently keep up. This causes high query latency while performing bulk imports. When write throughput decreases (<= 100 per second) the indexer catches up, and queries will be fast.
Most use-cases do an initial bulk import, followed by queries with lower write throughput (<= 100 per second). For this use-case, it's not a problem. We are actively working to improve this limitation.
There isn't a limit or performance metric we don't think we can improve by an order of magnitude when prioritized! If you expect to brush up against a limit or are limited by performance by an operation, contact us. Often can be fixed in days.
Metric | Max seen in production | Production limits (current) | Production limits (soon) |
---|---|---|---|
Max vectors (global) | 5B+ | ||
Max vectors (per namespace) | 100M+ | ||
Number of namespaces | 2M+ | Unlimited | |
Max dimensions | 10,752 | ||
Max inactive time in cache | Contact us if you need custom | ||
Write rate (global) | 10,000 | ||
Write rate (per namespace) | 10,000 | 10,000 vec/s | 100,000 vec/s |
Max write batch rate (per namespace) | 1 batch/s | ||
QPS (global) | ~100 QPS | ||
Max QPS (per namespace) | ~20 QPS | 100+ QPS | 10,000 QPS |
~90-95% | ~90-95% | Configurable | |
Max attribute value | 64 KiB | 1 MiB | |
Max attribute name length | 128 | 128 | 128 |
Max attributes per vector | 256 | 256 | 256 |
Max namespace name length | 128 | 128 | 128 |