Overview

Aeron is a framework for low-latency messaging and high-availability clustering, built for electronic trading and other systems where p99 latency matters more than averages.

Where to learn Aeron internals

This site is deliberately thin on internals. For a deep, authoritative explanation of how Aeron works under the hood — the transport protocol, log buffers, archive, and the Raft election state machine — read The Aeron Files. This guide assumes that knowledge and focuses on the operational side: tuning, measuring, monitoring, and recovering.

The three components

Layer	What it does	Builds on
Transport	UDP-based reliable, zero-copy messaging	— (standalone)
Archive	Durable recording & replay of streams	Transport
Cluster	Raft consensus for fault-tolerant stateful services	Transport + Archive

Transport is standalone. Cluster requires all three. (See The Aeron Files for the full architecture.)

How this guide is organized

• Performance Tuning — the levers that move latency and throughput.
• Benchmarking — reading benchmark numbers honestly, micro-bursts, and loss injection.
• Monitoring — metrics, alerting, and capturing hot-path data without perturbing it.
• Operations & Resilience — standby, DR, and the failure-mode runbook.
• OMS / ME Best Practices — sharding, rolling upgrades, and order-book design for an exchange on Aeron Cluster.
• News & Notes — field notes, and what changed recently on the site.

Design goals to keep in mind

Aeron exists to satisfy three non-negotiable requirements: ultra-low latency (single-digit µs), high throughput (millions of msg/s per stream), and predictable tail latency (p99/p999). Almost every tuning decision in this guide is a trade-off between these three.