Resilience - Spatial Data Management on AWS

Resilience

The AWS global infrastructure is built around AWS Regions and Availability Zones. AWS Regions provide multiple physically separated and isolated Availability Zones, which are connected with low-latency, high-throughput, and highly redundant networking.

Disaster Recovery

What the Solution Provides

The solution includes these built-in data protection features:

  • DynamoDB Point-in-Time Recovery (PITR): Continuous backups with 35-day retention for all metadata tables

  • S3 Versioning: Enabled on the asset bucket to protect against accidental deletion and overwrites

Recommended Best Practices

For production deployments, implement these additional backup measures:

  1. DynamoDB On-Demand Backups

    • Create on-demand backups before major changes (schema updates, bulk data operations, version upgrades)

    • On-demand backups are retained until explicitly deleted, unlike PITR’s 35-day limit

    • Use AWS Backup to schedule regular DynamoDB backups (daily or weekly)

  2. S3 Asset Bucket Backups

    • Configure AWS Backup to create regular backups of the asset bucket

    • Consider S3 Cross-Region Replication (CRR) to a secondary region for regional disaster recovery

  3. Cross-Account Backup (for critical data)

    • Copy DynamoDB backups to a separate AWS account for protection against account-level issues

    • Replicate critical S3 assets to a bucket in a different AWS account

For true disaster recovery from regional failures, implement cross-region replication and multi-region deployment.

High Availability

Multi-AZ Deployment

  • VPC spans 2 Availability Zones

  • NAT Gateways deployed in both AZs

  • Lambda functions can execute in either AZ

  • DynamoDB and S3 are inherently multi-AZ