Arpio protects Amazon Aurora MySQL and PostgreSQL clusters by copying incremental snapshots, or replicating in real time, to a separate AWS account and region. At recovery it rebuilds the cluster, its instances, and its reader, writer, and custom endpoints, along with the parameter groups, subnet groups, security groups, and IAM roles the cluster depends on.
Disaster Recovery for Amazon Aurora
What Arpio protects in Amazon Aurora
| Resource | What Arpio does |
|---|---|
| Aurora clusters (provisioned and Serverless) | Copies incremental snapshots or replicates in real time; rebuilds instances to match the primary |
| Reader, writer, and custom endpoints | Recreates them to match the original configuration |
| Cluster parameter groups | Copies them unchanged |
| DB subnet groups | Recreates them on the recovered VPC and subnets |
| Security groups and associated IAM roles | Selects and recovers them automatically |
| Encryption at rest | Creates a KMS key in the recovery environment; AWS re-encrypts snapshots with it during the copy |
| RDS Proxy | See RDS page |
Aurora Global Database vs. full Aurora disaster recovery
Aurora gives you snapshots, cross-region snapshot copy, and Aurora Global Database. Those protect the data. They don’t recreate the subnet groups, security groups, parameter groups, IAM roles, and endpoints a cluster needs in a new account, or bring up the application that connects to it. Arpio does both, and keeps hostnames working: it creates DNS aliases for reader and writer endpoints so applications reconnect without config changes.
How Aurora disaster recovery works with Arpio
- Pick a recovery point objective. Arpio copies incremental snapshots (only changed data moves, so a 10 MB change copies 10 MB) or replicates in real time.
- Arpio selects the cluster’s parameter groups, subnet groups, security groups, and IAM roles automatically.
- At recovery, Arpio builds the cluster and instances in compatible availability zones, spread across zones like the original.
- Endpoints come back to match, and DNS aliases let clients keep using their original hostnames.
- When you fail back, Arpio returns just the data to your original cluster, so your infrastructure-as-code and CI/CD pick up where they left off.
Testing and failback for Amazon API Gateway
Run a recovery test whenever you want. Arpio brings up the recovery environment in test mode while production keeps serving traffic, so you can confirm the application works without a maintenance window. When the outage is over, Arpio copies the latest data from the recovery environment back to your original infrastructure, and you can rehearse failback as often as you like.
Services recovered alongside Aurora
Amazon VPC, AWS KMS, AWS IAM, Amazon RDS (RDS Proxy), Amazon EC2, Amazon ECS, AWS Lambda.
Aurora requirements and limits
- Supported engines: Aurora MySQL and Aurora PostgreSQL.
- Aurora Global Database, Parallel Query, and Multi-Master clusters aren’t supported.
- Clusters recover with the engine’s default option group, since Aurora doesn’t support custom option groups.
Aurora disaster recovery FAQ
How is Arpio different from Aurora Global Database?
Global Database replicates one cluster to another region. Arpio recovers the cluster plus its network, security, and application dependencies into a separate account. Arpio doesn’t protect Global Database clusters today.
Does Arpio copy my whole database every time?
No. Aurora snapshots are incremental, so only changed data is copied.
Will my applications need new connection strings?
No. Arpio creates DNS aliases for the reader and writer endpoints.
Can Arpio protect encrypted Aurora clusters?
Yes. It provisions a KMS key in the recovery environment and the snapshot is re-encrypted during the copy.