Honua
// pillar 06 · high performance

AOT-compiled.
Benchmarked.
Cloud-optimized end to end.

Fast cold starts and a low memory floor from a NativeAOT-compiled engine. p95 latency tracked on every release against an open benchmark suite. Cloud-optimized formats and a CDN-friendly cache surface absorb the long tail — fast at any zoom, without hand-tuning a server pool.

Talk to us See benchmarks ↗
Workload shape · lower-is-betterillustrative
FeatureServer querycolumnar read
Legacy
baseline
Honua
faster
MapServer render · 1024 px tileraster path
Legacy
baseline
Honua
faster
Vector tile · pbfPMTiles range-fetch
Legacy
n / a
Honua
supported
Cold start · scale-to-zeroNativeAOT
Legacy
n / a
Honua
supported
// shapes are illustrative · methodology + p50 / p95 / p99 results live in geobench ↗ view methodology
// what makes it fast

Three things, none of them magic.

Performance is a property the engine is engineered for, not a setting. AOT compilation, an open benchmark suite the engine is tuned against, and cloud-optimized I/O all the way down.

01 · AOT-compiled engine Cold start measured in milliseconds, not seconds.

Honua Server is compiled ahead-of-time with .NET NativeAOT. No JIT warmup, low memory floor, container images small enough that serverless cold-starts are practical. The same binary runs in ECS, AKS, Lambda, or your laptop — with the same startup profile.

02 · Benchmarked & optimized An open benchmark suite that the engine is tuned against.

geobench is public: representative workloads, repeatable harness, p50 / p95 / p99 / RPS / err / cold-start tracked per release. When a hot path regresses, we see it in CI before you do — and the fix ships with a number, not a vibe.

03 · Cloud-optimized end to end Range-fetched data, CDN-cacheable output.

GeoParquet for features, Cloud-Optimized GeoTIFF for rasters, PMTiles + vector tiles for the map — range-fetched directly so a query reads bytes, not files. Tile and feature responses are cacheable URLs with strong ETags; drop a CDN in front, watch p95 fall again.

// related pillars
04 Cloud-native Container-first deployment, serverless functions, and elastic scalability. Read pillar 04 03 Predictable cost Open-core engine, elastic runtimes, no surprise licensing. Read pillar 03 01 Compatibility Esri-compatible service surface, OGC standards, and SDKs. Read pillar 01
// next

Run the benchmarks
against your own workload.

Talk to us See methodology ↗