A checkout saga spans inventory, payment, shipping, and loyalty. Downstream latency shifts every hour. Black Friday is not the day to discover your payment step timeout is baked into application.yml across twelve Spring Boot services.

Kiponos.io gives every saga participant the same live orchestration parameters — step timeouts, retry budgets, compensation triggers — via one shared config tree. Each JVM reads locally on every saga step; ops adjusts once in the dashboard; WebSocket deltas propagate without redeploying the fleet.

Why sagas break with static config

Typical saga coordinator code:

if (step.elapsedMs() > 8000) {
    compensate("payment", sagaId);
}

That 8000 usually comes from:

1. Per-service YAML — payment service says 8s, inventory says 12s; nobody agrees during an incident
2. Env vars in Helm — change means rolling twelve deployments
3. Shared DB config table — poll per step adds latency and coupling

Saga steps are high-frequency reads inside workflow engines. You need local memory reads and async updates — the same contract as live API rate limits.

Architecture: one tree, many participants

┌─────────────────┐     WebSocket deltas      ┌──────────────────────┐
│  Kiponos.io UI  │ ────────────────────────► │  Inventory service   │
│  platform ops   │                           │  Payment service     │
└─────────────────┘                           │  Shipping service    │
                                              │  (each: in-mem SDK)  │
                                              └──────────┬───────────┘
                                                         │ .getInt() local
                                                         ▼
                                              ┌──────────────────────┐
                                              │  saga step executor  │
                                              └──────────────────────┘

Every participant connects to profile ['orders']['v2']['prod']['sagas']. When NOC extends payment.step_timeout_ms, all JVMs see the new value on the next step — no config server poll, no inter-service "what is timeout now?" REST calls.

Shared saga config tree

sagas/
  checkout/
    payment/
      step_timeout_ms: 8000
      max_retries: 2
      retry_backoff_ms: 500
      compensate_on_timeout: true
    inventory/
      step_timeout_ms: 5000
      max_retries: 3
      hold_ttl_seconds: 120
    shipping/
      step_timeout_ms: 12000
      fallback_carrier: ups_ground
    global/
      saga_ttl_minutes: 30
      alert_on_compensation: true

Platform ops edits one folder; payment, inventory, and shipping services each read their subtree locally.

Java integration (saga participant)

import io.kiponos.sdk.Kiponos;

@Component
public class PaymentSagaStep {
    private final Kiponos kiponos = Kiponos.createForCurrentTeam();

    public StepResult execute(SagaContext ctx) {
        var cfg = kiponos.path("sagas", "checkout", "payment");
        int timeoutMs = cfg.getInt("step_timeout_ms");
        int maxRetries = cfg.getInt("max_retries");

        return withTimeout(timeoutMs, () -> capturePayment(ctx))
            .onTimeout(() -> cfg.getBool("compensate_on_timeout")
                ? compensate(ctx) : StepResult.retry(maxRetries));
    }
}

getInt() is a local cache lookup — safe inside the saga executor hot path.

Optional audit when ops changes timeouts mid-incident:

kiponos.afterValueChanged(change ->
    log.warn("Saga config changed: {} → {}", change.path(), change.newValue())
);

Real-world scenarios

Compensation policy without redeploy

Compensation is not just timeouts — trigger thresholds can live in the same tree:

boolean shouldCompensate = kiponos.path("sagas", "checkout", "global")
    .getBool("alert_on_compensation");
int sagaTtl = kiponos.path("sagas", "checkout", "global")
    .getInt("saga_ttl_minutes");

Risk and ops teams tune how aggressive the saga is while traffic is live.

Performance

• One WebSocket per JVM — not a config fetch per saga step
• Reads are O(1) on the SDK cache — microseconds per step
• Delta patches — changing one timeout does not reload the full tree
• No DB poll on the workflow hot path

Compare to alternatives

Getting started

1. Free TeamPro at kiponos.io — one profile for sagas/checkout/*
2. Add io.kiponos:sdk-boot-3 to each saga participant
3. Wire KIPONOS_ID, KIPONOS_ACCESS, and -Dkiponos=... on every service
4. Replace hard-coded timeouts with kiponos.path("sagas", ...).getInt(...)
5. Run a chaos test — slow payment mock, extend timeout in dashboard, watch compensations stop misfiring

Runnable golden example and Agent Skills: github.com/kiponos-io/kiponos-io

What is next

Sagas share state with handoff signals and event routing rules — other microservices patterns in the same live tree: who owns the lock, which topic fires next, when to escalate to manual review.

Kiponos.io — real-time config for Java. Tune distributed sagas while orders are in flight.