# Data Model — Spec 091 (BackupSchedule Retention & Lifecycle)

## Entities

### BackupSchedule (existing)

**Table**: `backup_schedules`

**Existing fields (selected)**
- `id` (pk)
- `tenant_id` (fk → `tenants.id`)
- `name` (string)
- `is_enabled` (bool)
- `timezone` (string)
- `frequency` (`daily|weekly`)
- `time_of_day` (time)
- `days_of_week` (json nullable)
- `policy_types` (json)
- `include_foundations` (bool)
- `retention_keep_last` (int)
- `last_run_at` (datetime nullable)
- `last_run_status` (string nullable)
- `next_run_at` (datetime nullable)
- `created_at`, `updated_at`

**New fields (this spec)**
- `deleted_at` (datetime nullable)

**State**
- Active: `deleted_at = null`
- Archived: `deleted_at != null`

**Validation / rules (lifecycle)**
- Archive is allowed only for active schedules.
- Restore is allowed only for archived schedules.
- Force delete is allowed only for archived schedules.
- Force delete is blocked if historical runs exist.

**Relationships (existing)**
- `tenant(): BelongsTo`
- `operationRuns(): HasMany` (derived from `operation_runs` using `tenant_id` + `context->backup_schedule_id`)

## Operational Records

### OperationRun (existing)

Historical runs for backup schedules are represented by `operation_runs` filtered via `BackupSchedule::operationRuns()`.

**Force delete constraint source**
- “Historical runs exist” is defined as: `BackupSchedule::operationRuns()->exists()`.

## Indexing / performance notes

- Add an index supporting common list queries for schedules:
  - at minimum, `backup_schedules(deleted_at)` for soft delete scoping
  - consider composite index like `backup_schedules(tenant_id, deleted_at)` if query planner needs it

No other schema changes are required for this feature.