Performance Tuning
The levers that move p50, p99, and throughput — BDP and receive-path buffers, NAK timers, batching and idle strategy, core pinning, NUMA, and when the DPDK media driver pays off.
LowLatencyOps
A field guide to running Aeron® in production — tuning, benchmarking, monitoring, and surviving the bad days.
The five pillars
Section titled “The five pillars”Benchmarking
How to read published Aeron numbers without being fooled by the burst knob, why steady-state TPS misses micro-bursts, and how injected loss moves p50 and p99.
Monitoring
Getting counters out of the CnC file into Prometheus or Datadog, the alerts to wire up, and capturing hot-path latencies without adding jitter.
Operations & Resilience
Cluster Standby, disaster recovery, backup strategies, and a failure-mode runbook: node, network, disk, snapshot, and human-error scenarios.
OMS / ME Best Practices
Building an exchange on Aeron Cluster: sharding the OMS by user and the matching engine by symbol, rolling upgrades, and choosing an order-book data structure.
Interactive tools
Section titled “Interactive tools” Buffer Sizing Calculator Enter your TPS, message size, and RTT; get copy-ready term buffer, window, SO_RCVBUF, and NAK timer values.
Step-down timeline Where every millisecond of a planned leader handoff goes — 266 ms stock down to ~57 ms, measured on AWS.
Cross-region migration walkthrough Step through node-by-node vs whole-cluster swap for moving a live cluster between regions.
New here?
Section titled “New here?” Read the 90-second overview What Aeron is, the three components, and where this site fits.
Follow the tuning methodology Sensible defaults, then incremental load: watch which metric degrades first and turn the one knob that addresses it.
Jump to the parameter reference Every transport knob and exactly how it pulls p50, p99, and throughput.
Read a benchmark honestly What the published cluster benchmark measures, and how to turn it into an EC2 sizing decision.
Recently updated
Section titled “Recently updated”- Performance Tuning Aeron® Cluster Snapshot Performance: A Matching-Engine Case Study From a snapshot that failed above 16 resting orders to a 5M-order take in 179 ms under ZGC, then a 3-node cluster where the Archive's per-block fsync, not encoding, sets the client pause.
- Operations & Resilience Tuning Cluster Failover Time How to reason about Aeron® Cluster leader failover time — the detection + election budget, which timeouts to turn, and the trade-off between fast recovery and self-inflicted false failovers. With measured results and a recommended sweet spot.
- Performance Tuning NAK Timer Tuning (Lever B) How to derive aeron.nak.multicast.max.backoff, group.size, and retransmit linger from RTT and receiver count — with measured results: a 79× tail collapse under seeded loss.
- Performance Tuning Bandwidth Delay Product (BDP) How BDP determines minimum buffer sizing for sustained throughput across CPG, same-AZ, cross-AZ, and cross-region round-trip latencies.
- Performance Tuning The Four Receive-Path Buffers The four nested buffers a packet traverses on the receive path — ENA RX ring, OS socket buffer, Aeron term buffer, and the flow-control window — their constraints, and how to size them as a consistent stack.
- Operations & Resilience Archive Retention and Purge: How Much Log Can You Delete? The position algebra that decides whether a node can restart or must be rebuilt, the two retention models to choose between, and a detach-validate-delete purge workflow that keeps every recovery path intact.
This site is not affiliated with, endorsed by, or sponsored by Adaptive Financial Consulting Limited or the Aeron project. Aeron is a registered trademark of Adaptive Financial Consulting Limited.
Aeron is a trademark of Adaptive Financial Consulting Limited in the United Kingdom and other countries.