Benchmarks

bunqueue includes a comprehensive benchmark suite for performance testing, stress testing, and production validation.

Quick Results

Push Throughput

1.2M+ ops/sec

With batch operations (1000+ jobs)

Process Rate

494,000 ops/sec

Full cycle with 16 workers

Scale Tested

1 Million Jobs

Verified data integrity at scale

Memory Efficient

~100 bytes/job

At scale with removeOnComplete: true

Live Benchmark Results

These results were measured on Mac Studio M1 Max, 32GB RAM, Bun 1.3.8.

Throughput by Operation

Operation	Throughput	Notes
PUSH	579,603 ops/sec	Single job insertion
PULL	420,835 ops/sec	Single job retrieval
ACK	464,857 ops/sec	Single job acknowledgment
BATCH(100)	1,259,234 ops/sec	Batch push, 100 jobs
BATCH(1000)	1,452,193 ops/sec	Batch push, 1000 jobs

Batch Size Scaling

Batch Size	Throughput	Speedup vs Individual
Individual	579,603/s	1x
100	1,259,234/s	2.2x
1,000	1,452,193/s	2.5x
5,000	1,393,745/s	2.4x

Worker Benchmark (Realistic Workload)

Scale	Push Rate	Process Rate	Total Events
10,000 jobs	519,445/s	217,980/s	✅ 10,000
50,000 jobs	604,918/s	243,945/s	✅ 50,000
100,000 jobs	611,819/s	232,613/s	✅ 100,000

Million Jobs Benchmark

Metric	Result
Total Jobs	1,000,000
Push Rate	1,210,654 ops/sec
Process Rate	494,805 ops/sec
Overall Rate	351,247 ops/sec
Total Time	2.85 seconds
Data Integrity	✅ PASSED (100%)

Stress Test Results

Test	Configuration	Throughput	Status
High Volume	100K jobs, single queue	140,032/s	✅ PASS
Concurrent Queues	10 queues × 10K jobs	138,336/s	✅ PASS
Large Payloads	5K jobs × 10KB each	384,781 MB/s	✅ PASS
Priority Stress	50K jobs, random priorities	179,983/s	✅ PASS
Retry Storm	10K jobs, 50% fail rate	-	✅ PASS
Batch Operations	100K jobs, batch=1000	266,619/s	✅ PASS

Benchmark Suite Overview

bunqueue includes 6 specialized benchmarks:

Benchmark	Purpose	Jobs	Duration
`throughput.bench.ts`	Individual operation speeds	10K-50K	~30s
`worker.bench.ts`	Realistic worker simulation	10K-100K	~60s
`stress.bench.ts`	Production validation (7 tests)	5K-100K	~3min
`million-jobs.bench.ts`	High-volume integrity test	1M	~5min
`10million-jobs.bench.ts`	Extreme stress test	10M	~15min
`completed-events.bench.ts`	Event delivery verification	10K	~30s

Running Benchmarks

# Clone and install
git clone https://github.com/egeominotti/bunqueue.git
cd bunqueue
bun install

# Individual benchmarks
bun run src/benchmark/throughput.bench.ts
bun run src/benchmark/worker.bench.ts
bun run src/benchmark/stress.bench.ts
bun run src/benchmark/million-jobs.bench.ts
bun run src/benchmark/10million-jobs.bench.ts
bun run src/benchmark/completed-events.bench.ts

Detailed Benchmark Documentation

File: src/benchmark/throughput.bench.ts

Measures raw throughput of individual queue operations with warmup.

Operations Tested

Operation	Description	Measured Speed
`PUSH`	Single job insertion	579,603/s
`PULL`	Single job retrieval	420,835/s
`ACK`	Single job acknowledgment	464,857/s
`FULL CYCLE`	Push → Pull → Ack	182,822/s
`BATCH PUSH (100)`	Batch insertion	1,259,234/s
`BATCH PUSH (1000)`	Batch insertion	1,452,193/s

How It Works

// 1. Warmup (100 iterations)
for (let i = 0; i < 100; i++) {
  await fn();
}

// 2. Timed execution
const start = performance.now();
for (let i = 0; i < totalJobs; i++) {
  await fn();
}
const durationMs = performance.now() - start;

// 3. Calculate throughput
const jobsPerSecond = (totalJobs / durationMs) * 1000;

File: src/benchmark/worker.bench.ts

Simulates realistic worker behavior with event verification.

Test Flow

┌─────────────────┐     ┌─────────────────┐     ┌─────────────────┐
│   Push Phase    │────▶│  Process Phase  │────▶│  Event Check    │
│                 │     │                 │     │                 │
│ • Push N jobs   │     │ • Pull job      │     │ • Count events  │
│ • Sequential    │     │ • Ack with data │     │ • Verify count  │
└─────────────────┘     └─────────────────┘     └─────────────────┘

Results

Size	Push Rate	Process Rate	Events
10,000	519,445/s	217,980/s	✅ 10,000
50,000	604,918/s	243,945/s	✅ 50,000
100,000	611,819/s	232,613/s	✅ 100,000

File: src/benchmark/stress.bench.ts

Comprehensive production validation with 7 different stress scenarios.

Test Suite

Test	Configuration	Result
High Volume	100K jobs	140,032/s ✅
Concurrent Queues	10 × 10K	138,336/s ✅
Large Payloads	5K × 10KB	384,781 MB/s ✅
Priority Stress	50K random	0 violations ✅
Retry Storm	10K, 50% fail	75% retry ratio ✅
Memory Stability	100 × 1K	Stable ✅
Batch Operations	100K batch	1,393,745/s push ✅

File: src/benchmark/million-jobs.bench.ts

High-volume test with data integrity verification.

Configuration

Parameter	Value
Total Jobs	1,000,000
Batch Size (push)	5,000
Batch Size (pull)	500
Workers	16
Queues	16

Results

📊 Summary
==========================
Total jobs:      1,000,000
Completed:       1,000,000
Total time:      2.85s

Push rate:       1,210,654 jobs/sec
Process rate:    494,805 jobs/sec
Overall rate:    351,247 jobs/sec

✅ Data Integrity: PASSED
   - Pull: 1,000,000 / 1,000,000 unique
   - Events: 1,000,000 / 1,000,000 unique

File: src/benchmark/10million-jobs.bench.ts

Extreme stress test to validate system limits.

Configuration

Parameter	Value
Total Jobs	10,000,000
Workers	32
Queues	32
Shards	32

Visual Progress

👷 Processing 10M jobs...
[██████████████░░░░░░░░░░░░░░░░] 45.2% | 428,571 jobs/sec
[████████████████████░░░░░░░░░░] 67.8% | 435,897 jobs/sec
[██████████████████████████████] 100.0% | 441,176 jobs/sec

File: src/benchmark/completed-events.bench.ts

Verifies that every COMPLETED event is reliably emitted.

Results

📊 RESULTS
══════════════════════════════════════════════════
Total jobs:         10,000
COMPLETED events:   10,000
Unique job IDs:     10,000
──────────────────────────────────────────────────
Push time:          20ms (506,778 jobs/sec)
Process time:       47ms (211,589 jobs/sec)
──────────────────────────────────────────────────
✅ TEST PASSED: All COMPLETED events received!

Performance Tuning Guide

Batch Size Recommendations

Scenario	Push Batch	Pull Batch
Low latency (real-time)	100-500	10-50
High throughput	1,000-5,000	500-1,000
Extreme volume	5,000-10,000	1,000+

Worker Count Tuning

Optimal Workers = min(CPU cores, shard count)

bunqueue uses 16 shards by default:

4-core machine: 4-8 workers
8-core machine: 8-16 workers
16+ core machine: 16 workers

Memory Optimization

removeOnComplete: true

Maximum throughput (~495K/s)

Jobs removed after completion
No result storage
Best for fire-and-forget

removeOnComplete: false

~25% slower (~370K/s)

Jobs tracked after completion
Results queryable
Required for dependencies

Why bunqueue is Fast

Native SQLite

Direct FFI bindings via bun:sqlite. WAL mode for concurrent reads/writes. Memory-mapped I/O.

16 Shards

Minimizes lock contention. Each shard handles ~6% of queues. Parallel processing by design.

Batch Optimization

Single transaction for bulk operations. Prepared statements reused. ackBatchWithResults for throughput.

Efficient Structures

Skip lists for O(log n) priority. MinHeap for delayed jobs. Bounded collections prevent leaks.

Benchmark Environment

For reproducible results, note your environment:

# System info
uname -a
bun --version

# CPU info (macOS)
sysctl -n machdep.cpu.brand_string

# CPU info (Linux)
cat /proc/cpuinfo | grep "model name" | head -1

Reference hardware (used for these results):

Mac Studio, Apple M1 Max, 32GB RAM
macOS 26.2 (Tahoe)
Bun 1.3.8
SSD storage

Pass Criteria Summary

Benchmark	Pass Condition	Result
Throughput	Operations complete	✅
Worker	`events === jobs`	✅
Stress: High Volume	`events === jobs`	✅
Stress: Concurrent	`events === totalJobs`	✅
Stress: Priority	`orderViolations === 0`	✅
Stress: Retry	`completed + dlq === jobs`	✅
Million Jobs	Data integrity 100%	✅
Events	`completedEvents === jobs`	✅

Submit Your Results

Run benchmarks on your hardware and share results via GitHub Discussions to help others compare.

Include:

CPU model
RAM
Bun version
OS
Benchmark output