# Operational requirements

Product design

This page describes the product model and requirements. Examples illustrate the model, not a guaranteed API or completed feature. Check the [application contracts](/stormtrust/apps/) and [dataset inventory](/stormtrust/product/platform-and-data/datasets-and-source-contracts/) for implemented scope. Use [operations guides](/stormtrust/engineering/operations/) for runnable procedures.

StormTrust deployments are reviewable, reproducible, observable, and reversible when possible. Terraform manages durable infrastructure. Wrangler manages application deployments.

For supported procedures, use [Deploy a change](/stormtrust/engineering/operations/deploy/), [Check application health](/stormtrust/engineering/operations/health/), and [Recover a release](/stormtrust/engineering/operations/recover/). The following signals and controls describe the target operating model; their complete coverage is not implemented.

## Observability

### Application signals

Application signals include request count, errors, latency, CPU, and exceptions.

### Weather signals

Weather signals include source and product lag, missing frames, decoding failures, and event-derivation lag.

### Queue signals

Queue signals include backlog, oldest message, retries, and failures.

### Product signals

Product signals include active storms, generated alerts, delivery status, property impacts, and report failures.

### Cost signals

Cost signals include Workers, D1, R2, Queues, Durable Objects, AI, and external APIs.

## Source health dashboard

The target source dashboard exposes status, latest expected and received objects, lag, error rate, missing partitions, and last successful decode.

## Reprocessing

The target reprocessing interface supports a source asset, product, time range, event, revision, algorithm release, or region. Reprocessing never silently overwrites historical outputs.

## Incidents

Potential incident categories include:

| Severity | Definition                                                        |
| -------- | ----------------------------------------------------------------- |
| SEV-1    | System-wide inability to provide critical live storm intelligence |
| SEV-2    | Material degradation of a key source or product                   |
| SEV-3    | Limited feature degradation                                       |

Exact definitions require operational approval before production use.

## Cost management

Free source data can still create compute, storage, request, database, queue, and rendering costs. The target cost model measures cost per source asset, event, tenant, report, integration, and AI task.

## Infrastructure recovery

For current infrastructure state recovery, refer to [State and drift](/stormtrust/engineering/cloudflare/state-and-drift/).