
What Businesses Build on TYO MQ
A message queue is infrastructure, and infrastructure is only interesting for what it lets you ship. This page walks through the applications businesses build on TYO MQ — and because claims are cheap, every use case here maps to a working sample in the tyo-mq-samples repository that you can run on your own machine in about a minute.
1. Webhook delivery your customers can trust
If your product promises to notify customer systems — payment events, order updates, sync triggers — you own a hard problem: their endpoint will go down, and it will be your support ticket. Building delivery with retries, backoff, and failure visibility from scratch is weeks of undifferentiated work.
On TYO MQ it's a subscription: each customer endpoint gets a durable queue; a courier POSTs the event and acknowledges only on success. The server holds unacknowledged events, retries on your schedule with exponential backoff, and moves permanently failing deliveries to a dead-letter queue with the reason attached — where support can inspect and replay them once the customer fixes their server. Your platform never notices a customer's outage.
npm run webhook-delivery in the samples repo simulates a healthy
customer, one that recovers mid-retry, and one that stays dead — and shows
the dead-letter queue and Prometheus metrics at the end.
2. Background jobs and worker pools
Resize the images, generate the report, send the receipts — after the request, not during it. Subscribe a pool of workers with a consumer group and TYO MQ load-balances jobs across them round-robin; add a worker and throughput grows, lose one and the others carry on. Durable subscriptions mean jobs dispatched while a worker is down are replayed when it returns, and a poison job that keeps crashing its handler is retried, then dead-lettered — visible to an operator, not lost in a log file.
3. Live operations dashboards
Polling every thirty seconds is how dashboards lie to you. With TYO MQ, the
systems doing the work publish events to hierarchical topics —
orders/sydney/A4172/status — and a browser dashboard subscribes once with
a wildcard (orders/#) to see everything, live. The bundled browser client
means no separate push service and no per-message fees; the dashboard is just
another subscriber.
4. IoT and device fleets
Devices publish telemetry to their own topics
(org/site/device/telemetry) and listen for commands on another — the classic
MQTT shape. Operations subscribes with wildcards (org/+/+/telemetry) to
watch an entire fleet through one subscription, and sends targeted commands
back to any single device. Add sites and devices freely: the topic tree is the
whole contract, and nothing else changes.
5. Microservices that stay decoupled
Point-to-point service calls quietly become a web where nothing can be
deployed without breaking something else. With an event backbone, your order
service announces order-placed and doesn't know or care who listens —
inventory reserves stock, email sends confirmations, and next quarter's
analytics service starts listening without a single upstream change.
Consumer groups scale any service horizontally; durable subscriptions let a
service be redeployed without missing events. And the services don't have to
share a language: with clients for Node.js, Python, Go, Rust, C/C++, Java,
C#, and Ruby, each team builds in its own stack against the same backbone.
6. Multi-tenant SaaS event hubs
If your platform serves many customer organisations, TYO MQ's realms give each one an isolated event space on the same server — identical event names, zero crosstalk, enforced by token authentication rather than developer discipline. New customer integrations onboard through a signed approval workflow: the integration submits a request with its own token, your operator approves it, and the token starts working immediately, scoped to one realm. Revocation is one command.
7. Audit trails and compliance
Regulated businesses must answer "what happened, and when?" even for the
moments their systems were down. One durable wildcard subscription (#)
records every event in the realm; with SQLite or Redis persistence the
queue survives not just the audit service restarting, but the message server
itself being killed and restarted — the samples repo proves it by doing
exactly that, live.
One backbone, not seven systems
The quiet punchline: these aren't seven products. The webhook courier, the worker pool, the dashboard, the device fleet, the audit trail — they're all the same three ideas (publish, subscribe, durable when it matters) composed differently. That's what a backbone buys you: each new use case is an afternoon, not a procurement cycle.
Building something that isn't on this list? Tell us about it — TYO MQ is our own backbone, and we like hard messaging problems.
