Metadata Cleanup

The Metadata Cleanup project (Alert.CA.MetadataCleanup) is an Azure Functions v4 app that periodically purges expired image metadata from Table Storage. Without this service, metadata would accumulate indefinitely as cameras continuously capture images.

Architecture

How It Works

Step 1: Timer Enqueue

A timer-triggered function runs on a configured schedule and enqueues device IDs to three separate cleanup queues — one per image type.

Step 2: Queue Processing

Three independent queue processors run in parallel:

• Single image metadata cleanup — Processes single-metadata-delete queue
• Panoramic image metadata cleanup — Processes pano-metadata-delete queue
• Pre-stitched image metadata cleanup — Processes prestitch-metadata-delete queue

Step 3: Batch Deletion

For each device ID dequeued:

1. Query the metadata index repository for expired records
2. Batch delete records from both the index and metadata tables
3. Records older than the configured retention period are purged

Repositories Used

Each image type has paired repositories:

Image Type	Metadata Repository	Index Repository
Single	ISingleImageMetadataRepository	ISingleImageMetadataIndexRepository
Panoramic	IPanoImageMetadataRepository	IPanoImageMetadataIndexRepository
Pre-Stitched	IPreStitchImageMetadataRepository	IPreStitchImageMetadataIndexRepository

Configuration

The MetadataCleanupOptions class controls:

Option	Description
Retention period	How long metadata is kept before deletion
Batch size	Number of records deleted per batch operation
Schedule	CRON expression for the timer trigger

Documentation Sections

• Cleanup Operations — Timer and queue-based deletion flow
• Data Structure — Dual-table design: device-partitioned metadata and time-partitioned index tables
• Scaling & Parallelism — Three-queue parallel processing
• Configuration — Retention settings and schedule