Benchmark Analysis: Fastro v1.1.2 Performance Leap
With the release of Fastro v1.1.2, we've introduced several architectural refinements aimed at minimizing framework overhead. Our latest benchmarks show a significant performance leap, bringing Fastro even closer to—and in some cases beyond—bare-metal Deno.serve throughput.
The Results
The benchmark was executed using k6. We compared Fastro v1.1.2 against a native Deno implementation of the same logic.
Note: Benchmark results may fluctuate depending on system resources and load during execution.
| Scenario | Framework | Throughput (req/s) | Avg Latency | P95 Latency | % of Native |
|---|---|---|---|---|---|
| Root | Native | 67294.34 | 1.4ms | 2.45ms | 100% |
| Fastro | 64549.85 | 1.46ms | 2.54ms | 95.92% | |
| URL Params | Native | 58381.92 | 1.62ms | 2.78ms | 100% |
| Fastro | 61323.13 | 1.54ms | 2.62ms | 105.04% | |
| Query Params | Native | 50334.73 | 1.9ms | 2.5ms | 100% |
| Fastro | 54541.10 | 1.74ms | 2.51ms | 108.36% | |
| Middleware | Native | 60465.19 | 1.56ms | 2.63ms | 100% |
| Fastro | 61987.21 | 1.52ms | 2.62ms | 102.52% | |
| JSON POST | Native | 39613.52 | 2.4ms | 3.66ms | 100% |
| Fastro | 40514.39 | 2.34ms | 3.69ms | 102.27% |
Key Takeaways
1. Optimized Middleware Engine (~102.52%)
The most dramatic improvement is in the Middleware scenario. By moving from a dynamic per-request dispatch to Pre-compiled Middleware Chains, we've eliminated array allocation and concatenation overhead from the request path. At 102.52% of native speed, Fastro actually outperforms simple native implementations by using highly-optimized call structures that V8 can inline effectively.
2. V8 Hidden Class Stability with FastContext
We introduced the FastContext class to serve as the request context. This ensures that the context object has a stable "Hidden Class" (or Map) in V8. When middlewares add properties to the context, V8 can optimize these accesses much better than with plain object literals. This is a key reason for our high performance in scenarios with multiple middlewares.
3. Balanced Performance across all Scenarios
Fastro reaches 95% to 108% of native Deno throughput across all tested scenarios. This isn't just a win for raw speed; it's a win for reliability too. Along with these performance gains, we've achieved 100% project-wide test coverage, ensuring that the engine is as stable as it is fast.
The Architecture of Speed
These results are the fruit of technical optimizations in the v1.1.x series:
- Pre-compiled Chains: We compute the final middleware stack for every route once during
app.serve(). - FastContext: A dedicated class for request context to optimize property access and memory allocation.
- Eager Query Parsing: Streamlined parsing logic that ensures 100% test coverage without compromising speed in real-world scenarios.
- Unified Fast-Path: Merged global and route-level cache logic for zero-overhead routing.
Conclusion
Fastro v1.1.2 proves that you don't have to choose between a developer-friendly API and raw performance. With an engine that often exceeds native speeds and 100% test coverage, Fastro is the premier choice for high-performance Deno applications.
Benchmarks were performed on Sat Mar 07 2026. For more details on running your own benchmarks, check out our GitHub repository and BENCHMARK.md.