Benchmarks · Key-value · Seven engines · Over the wire
Rostam is two engines, so the KV half gets measured the same way as the vector half — seven engines over the wire, one machine, one session, with the caveats printed as large as the wins.
Raw GET throughput and p99 tails against Redis, Memcached, Dragonfly, Aerospike, Valkey and KeyDB — then replicated RF=2 writes at matched commit semantics. A self-hosted key-value store in Go — the second engine in the same Go module as the vector engine.
The numbers
Raw GET throughput first — one machine, one session. Replicated writes below.
| GET ops/s | Rostam | Memcached | Dragonfly | Aerospike | Valkey | Redis | KeyDB |
|---|---|---|---|---|---|---|---|
| 64 conns | 726.1k | 682.9k | 532.0k | 510.6k | 239.1k | 228.3k | 206.3k |
| 256 conns | 732.4k | 681.7k | 551.6k | 516.9k | 227.9k | 219.6k | 208.8k |
p99 at 512 connections: Rostam 2.02 ms, best of the seven (Memcached 2.32, Aerospike 2.75, Redis 4.69). Memcached wins at 8 connections; Rostam pulls ahead as concurrency rises.
RF=2 against Aerospike CE 8.1 — both acking only after the replica has the write.
| RF=2, replica-ack | 8 conns | 32 conns | 128 conns | p50 @128 |
|---|---|---|---|---|
| Rostam PB | 35.6k | 80.9k | 113.6k | 0.99 ms |
| Aerospike CE | 38.0k | 57.0k | 85.9k | 1.35 ms |
Redis 7 (WAIT 1) |
28.7k | 43.2k | 42.9k | 2.95 ms |
Valkey 8 (WAIT 1) |
28.0k | 42.2k | 45.9k | 2.72 ms |
Brass marks the per-column winner — which is why Aerospike keeps the 8-connection cell.
Methodology
The bounds on the key-value numbers, printed as large as the wins.
Aerospike wins at 8 connections — and has the better tail there — while Rostam overtakes at 32 and leads +42% / +32% at 32/128, with the better median from 32 upward. The Redis-protocol engines flatten above 32 connections, which is a per-instance ceiling.
Measured twice on the same box four days apart; Rostam reproduced within ~1–4% at the anchor points. Over a real network on separate nodes it tightens to near-parity, with Rostam's tail still ahead (p99 ~8 ms vs 17–22 ms).
Both runs, the full posture matrix, and why a single load generator invalidates this comparison entirely are in the open benchmark repository. And Rostam's other half — the vector engine — is measured with the same discipline on the vector benchmark page.
curl -fsSL https://rostamlabs.com/install.sh | sh