WJb Performance Evolution
Benchmark: 1,000,000 jobs, single worker, in-memory storage
| Stage | Jobs/sec | Gain vs Baseline |
|---|---|---|
| Baseline (.NET 10) | 181,697 | - |
| .NET 11 + InMemoryStore + StoreBase | 206,070 | +13.4% |
| JobExecute Update | 211,188 | +16.2% |
| WasmWorker Update | 213,723 | +17.6% |
Current WJb Result
Completed : 1,000,000
Expected : 1,000,000
Sum : 499,999,500,000
ExpectedSum : 499,999,500,000
Verification: PASSED
Elapsed : 00:00:04.6789558
Jobs/sec : 213,723
Memory : 993.87 MB
Competitors
Quartz
Completed : 1,000,000
Expected : 1,000,000
Sum : 499,999,500,000
ExpectedSum : 499,999,500,000
Verification: PASSED
Elapsed : 00:00:24.8474230
Jobs/sec : 40,246
Memory : 1,492.51 MB
Hangfire
Completed : 1,000,000
Expected : 1,000,000
Sum : 499,999,500,000
ExpectedSum : 499,999,500,000
Verification: PASSED
Elapsed : 00:05:55.4745246
Jobs/sec : 2,813
Memory : 2,527.67 MB
Comparison
WJb vs Quartz : 5.31x faster
WJb vs Hangfire : 75.98x faster
Memory Usage
WJb : 993.87 MB
Quartz : 1,492.51 MB
Hangfire : 2,527.67 MB
Performance Chart
Source Code
Don't trust benchmarks. Verify them.
All benchmark source code is available here:
https://github.com/UkrGuru/WJb.Demo/tree/main/benchmark/LargeScale
Included benchmarks:
- WJb
- Quartz
- Hangfire
Clone it, run it, and compare the results yourself.
Notes
WJb processed 1,000,000 jobs in under 5 seconds.
Quartz completed the same workload in about 25 seconds.
Hangfire completed the same workload in nearly 6 minutes.
By the time Hangfire finished the benchmark, my coffee was already cold. ☕😄
Apparently, Hangfire treats one million jobs as a serious enterprise process.
