Stratum
SQL that branches and beats DuckDB on 35/46 1T benchmarks
Details
- External ID
- 47357141
- Source
- HN
- Company
- —
- Product
- Stratum
- Website domain
- datahike.io
- Launched
- March 12, 2026
- Cohort
- —
- Upvotes
- 12
- Upvotes percentile
- 0.6439114391143912
- Tags
- —
- Fetched at
- Sept. 7, 2026, 9:26 p.m.
- Updated at
- Sept. 7, 2026, 9:26 p.m.
Enrichment
- Theme
- database infrastructure and developer tools
- Vertical
- Horizontal
- Function
- Data infrastructure
- Audience
- Developer
- AI stance
- Not AI
- Project type
- Commercial product
- Normalized one-liner
- sql engine with branching
- Manually corrected
- False
Could you build this?
No Writing an analytical columnar database engine that implements copy-on-write branching, Java Vector API SIMD acceleration, and outperforms DuckDB is deep database systems research.
What it would actually take: The architecture consists of a custom columnar storage engine utilizing structural sharing (CoW trees) to allow zero-cost branching, combined with vectorized query execution utilizing the Java Incubator Vector API for SIMD instructions, and an embedded Postgres wire-protocol parser. The hard part is authoring custom vectorized SIMD kernels (filter, aggregate, join hash tables) and query execution planners that avoid JVM GC overhead and achieve hardware-saturating scan throughput.
Discussion
3 comments analyzed.
Competitors mentioned: DuckDB, Datahike, Proximum (HNSW vector index), Scriptum (Lucene)
Concerns raised: Columnar overhead for random access writes vs row-wise indices, Inefficient storage of fully variable strings in column format, Lack of benchmark comparisons against alternatives, Write throughput claims not yet validated in formal benchmarks
Feature requests: Better support for variable-length string storage, Online update optimization for random access write workloads
Competitors
Other products that read as similar to this one — 710 launches clear the similarity bar, closest 8 shown.
Attention rank: #253 of 711 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).
Launched 134 days after the earliest competitor.
- VaultSQL · hn · 2026-06-06 · 8 upvotes · similarity 0.62
- Real-workload SQLite benchmarks on Hetzner's cheapest VPS · hn · 2026-05-09 · 8 upvotes · similarity 0.61
- sqldoom · github · 2026-09-22 · 19 upvotes · similarity 0.59
- TidesDB · hn · 2025-11-09 · 14 upvotes · similarity 0.58
- jevql · github · 2026-09-18 · 11 upvotes · similarity 0.58
- LaminarDB · hn · 2026-02-18 · 6 upvotes · similarity 0.57
- Bun-sqlgen · hn · 2026-06-23 · 55 upvotes · similarity 0.57
- Real-Time, Streaming SQL Queries on Flight Data · hn · 2026-04-13 · 5 upvotes · similarity 0.57
Other launches for this product
- No other launches for this product.
Same idea, different domain
Nobody's really built a data infrastructure tool for Media & entertainment yet.