Khone docs

Khone is an HTTP microbatching gateway for AWS Lambda. It groups short-lived HTTP requests for a few milliseconds, invokes target Lambda functions with batched payloads, and returns each response to the right caller.

Use these docs to answer one question at a time: whether Khone fits your workload, how to deploy the gateway, how to connect handlers, how to tune it, and where to look up exact fields and protocols.

Start

When Khone helps: decide whether the latency-vs-cost tradeoff fits your workload.
Quickstart: deploy the bootstrap stack, deploy one example gateway, and send a request through it.

Deploy

LMI deployment model: understand the gateway function, bootstrap stack, Function URL, and Lambda Managed Instances shape.
Example templates: deploy one of the included SAM examples.
SAM gateway: adapt the deployment pattern to your own application stack.

Integrate

Choose an integration mode: pick adapters, native batch, or the layer proxy.
Adapters: wrap normal Node or Rust handlers.
Native batch: handle the full batch yourself when shared work matters.
Layer proxy: use an experimental compatibility path for unmodified handlers.

Operate

Tune batching: choose wait windows, batch sizes, timeout bounds, and isolation keys.
Observability: enable traces, EMF metrics, and debug headers.

Benchmarks

Benchmark results: read the public cost and latency snapshot.
Benchmark methodology: understand what the public reports measure.
Deploy the benchmark stack: create the dedicated benchmark environment.
Run benchmarks: run k6 and render report bundles.

Reference

Configuration: author GatewayConfig, Spec, and x-khone.
Batch protocol: implement target request and response payloads.
Streaming protocol: stream chunks per request from one batched invocation.
Bootstrap macro: publish gateway config artifacts from CloudFormation.
SDK adapters: look up adapter APIs.
Benchmark CLI: look up benchviz commands and flags.

Status

Khone is experimental. Interfaces may change while the LMI deployment model settles.