Honua
// modern cloud-native GIS

Run GIS in
your cloud.

Honua turns the GIS serving tier into familiar cloud infrastructure. Run the same OCI image with Docker or the public Helm chart, keep operational data in PostGIS, and send telemetry to the tools your platform team already uses. You choose the account, network, region, and scaling policy.

honua · deployed in your cloud
# serving tier
Docker · public Helm chart
ECS/Fargate · Lambda · Container Apps · Functions
Kubernetes (EKS/AKS)
 
# batch compute
AWS Batch · Azure Batch · Kubernetes Jobs
 
# data tier
PostGIS · S3 · Azure Blob · MinIO
 
# platform signals
health · readiness · OpenTelemetry
// familiar foundations

Modernize the serving tier.
Keep the infrastructure you trust.

Honua fits into a normal platform architecture, so GIS does not need its own isolated operating model.

01 · Containers first One portable unit across your environments.

The same OCI image works with Docker or the public Helm chart. Reference paths cover Kubernetes (EKS/AKS), AWS ECS/Fargate, AWS Lambda, Azure Container Apps, and Azure Functions, each with its own documented image family and rollout mechanism. During a pilot, validate production HA, sizing, upgrades, and rollback for your environment.

02 · Data stays yours Keep PostGIS as the operational system of record.

Connect the database you already secure, back up, and monitor. Store bounded artifacts in S3, Azure Blob, or MinIO, and serve supported COG and GeoParquet workflows without first moving every object into a proprietary store.

03 · Signals travel Bring GIS into your existing observability practice.

Health and readiness endpoints work with platform probes. OpenTelemetry exports traces, metrics, and structured logs to the backend you configure, giving GIS requests the same operational context as other services.

// scale on your terms

Add capacity where the workload needs it.

Start with one container, then add serving replicas as traffic grows. Multi-node deployments use shared storage and Redis so coordination remains explicit. Put finite tile and artifact keys behind your CDN, and size PostGIS independently for the data and query load.

Compute Scale the serving tier without relocating the system of record.
Storage Use object storage for durable artifacts and cloud-native datasets.
Operations Use the probes, telemetry, rollout controls, and cost tools your platform already provides.
// runtimes and batch compute

Pick the runtime that fits the traffic.
Send heavy jobs somewhere else.

Steady traffic and spiky traffic want different runtimes, and long-running geoprocessing wants neither. Honua separates the serving tier from batch compute so each one scales on its own terms. Infrastructure-as-code for these patterns ships as Terraform modules available to customers through support.

Managed runtimes Container or serverless, on either cloud.

Each pattern has its own verified image family and rollout mechanism — weighted target groups and alias versions for canaries on AWS, revision traffic splitting and atomic slot swaps on Azure. Serverless targets scale to zero for spiky or low-traffic workloads.

AWS ECS/Fargate
AWS Lambda
Azure Container Apps
Azure Functions
Kubernetes Fits an existing cluster estate.

The public Helm chart deploys the same multi-arch web image to EKS, AKS, or any conformant cluster. Serving replicas stay stateless, so scale-out is adding replicas — with Redis for durable jobs, imports, and workflows.

Amazon EKS
Azure AKS
Public Helm chart
Batch compute Geoprocessing runs off the request path.

Long-running geoprocessing and import jobs dispatch to a remote batch backend instead of the serving container. This matters most on serverless: a frozen or torn-down invocation cannot host a job, so Honua detects the substrate and routes work to batch rather than queuing jobs that cannot finish.

AWS Batch
Azure Batch
Kubernetes Jobs
// sizing

Runtime requirements and a measured datapoint.

Honua does not publish a requests-per-pod sizing rule. What exists publicly is one measured datapoint with a disclosed profile — a dated historical snapshot that cannot be independently reproduced exactly (the benchmark page states its own limits) — and the billing ceiling that bounds a serving unit. Use them as rough anchors, then validate against your own workload.

Measured datapoint · GeoBench, April 28, 2026 A 4-CPU-core / 4 GB container sustained the published per-scenario figures.

In the historical GeoBench snapshot, the Honua server container was limited to 4 CPU cores and 4 GB memory (PostGIS and the k6 load generator each ran in their own 4-core / 4 GB container), with no spatial response cache and disclosed bounded active-query/pool settings 6/6/3, against a 100K-point dataset. Within that profile it sustained 1632.8 req/s at p95 5.2 ms on the attribute-filter LIKE scenario, 937.2 req/s at p95 13.8 ms on GeoServices FeatureServer/query (medium bbox), and 1190.2 req/s at p99 201.4 ms under the 100-VU concurrent mixed workload.

Billing ceiling · not a sizing target One serving unit is capped at 8 vCPU and 32 GiB for licensing purposes.

On the pricing page, one serving unit is one production serving node up to 8 vCPU and 32 GiB RAM. That ceiling is a billing construct — the largest node that still counts as one unit — not a minimum requirement, a recommendation, or a performance envelope.

A datapoint, not a guarantee Validate sizing against your workload during a pilot.

The figures above are a measured datapoint under one specific, disclosed profile — not a sizing guarantee. Your data model, query mix, cache policy, and latency targets will move the numbers. The GeoBench harness ↗ is Apache-2.0: point it at your own dataset and query mix as a sizing tool, then confirm the result during a pilot.

# Illustrative only — mirrors the disclosed GeoBench container profile
# (4 CPU cores, 4 GB memory). Not a recommendation or a guarantee;
# validate against your own workload during a pilot.
# Helm values / Kubernetes container resources:
resources:
  requests:
    cpu: "4"
    memory: "4G"
  limits:
    cpu: "4"
    memory: "4G"
Not yet Honua Cloud and AWS or Azure Marketplace listings are not available. Today, deploy the self-managed server in infrastructure you control.
// your cloud, your operating model

Modern GIS infrastructure.
No separate platform to learn.