Benchmarks

All numbers in this document are measured on real hardware against real datasets with --release builds. Random-vector HNSW benchmarks are not reported — only SIFT-1M or equivalent ANN-benchmarks datasets.

Note

Every number is reproducible. Commands are shown inline. No synthetic or estimated numbers are published.

Hardware

All benchmarks run on AWS c6id.4xlarge:

CPU: Intel Xeon Platinum 8375C (Ice Lake), 16 vCPU @ 2.90 GHz
RAM: 30 GiB
Storage: 884 GB NVMe (io_uring backend)
OS: Ubuntu 24.04
Build: cargo build --release
Commit: 8567691d4f7859742c1e6cb54ba8c429ae36d297
Timestamp: 2026-05-10T21:51:55Z

HNSW on SIFT-1M

Dataset: SIFT-1M — 1,000,000 × 128-dim float32 vectors, 10,000 queries, pre-computed ground truth.

Source: ftp://ftp.irisa.fr/local/texmex/corpus/sift.tar.gz
SHA256: 92f1270c5e3a0cb46b89983e72b0511e4df065c31a9fa0276d8c9b1fca5bc81a

Build: M=16, ef_construction=200 → 66.2 s (15,114 vec/sec)

ef_search	recall@10	mean latency	p99 latency
10	0.762	57.6 µs	101 µs
50	0.959	158.1 µs	228 µs
100	0.983	266.5 µs	364 µs
200	0.990	459.4 µs	616 µs

bash

# Reproduce
cargo run --release -p galaxdb-benchmarks -- \
    --sift-dir /path/to/sift \
    --ef-sweep 10,50,100,200 \
    --output bench-results/sift_bench.json

Storage OLTP

1M rows, 16 threads, 60 s, NVMe storage:

Metric	GalaxDB	RocksDB	PostgreSQL 16
Write TPS	258,555	~80,000	~3,200
Read p50	3 µs	~180 µs	~95 µs
Read p99	47 µs	~500 µs	~300 µs
Write p99	377 µs	1–10 s*	—

*RocksDB without write pacing (vLSM, arXiv 2024)

Storage OLAP

10M rows, 16 threads, 60 s column scan:

Metric	GalaxDB	PostgreSQL 16
Scan throughput	4.49 GB/s	~0.9 GB/s
Zone-map skip rate	80%	N/A (heap scan)
OLTP p99 during scan	191 µs	—
HotSet evictions from scan	0	—

Crash Safety

All 7 chaos scenarios pass in <30 s total:

Scenario	Result	Time
Kill mid-flush → WAL replay	✅ 1,000 rows recovered, zero loss	8.79 s
Kill mid-compaction → old blocks intact	✅ 4,000 keys readable	0.02 s
Corrupt WAL record → replay stops	✅ Partial recovery, no corrupt data	1.81 s
Disk full → clean checkpoint	✅ Reserve freed, reads continue	0.01 s
Kill sidecar → backlog preserved	✅ 50 requests queued, drained on recovery	0.00 s
100 concurrent writers	✅ 100K writes, 0 duplicates, 0 missing	0.13 s
OLAP scan during OLTP	✅ 0 HotSet evictions, OLTP p99 unaffected	0.15 s

bash

# Reproduce crash safety tests
cargo run --release -p galaxdb-chaos-tests

Test Suite

v1 release test results on AWS c6id.4xlarge:

Rust unit tests	740 passed / 0 failed
Chaos scenarios	7 passed / 0 failed
Total chaos suite time	10.91 s

bash

# Run the full test suite
cargo test --workspace --exclude galaxdb-python --lib

# Run chaos tests
cargo run --release -p galaxdb-chaos-tests

Docker Compose

Contributing