Amazon Aurora DSQL is provided as a Preview service.
To learn more, see Betas and Previews
Resilience in Amazon Aurora DSQL
The AWS global infrastructure is built around AWS Regions and Availability Zones (AZ). AWS Regions provide multiple physically separated and isolated Availability Zones, which are connected with low-latency, high-throughput, and highly redundant networking. With Availability Zones, you can design and operate applications and databases that automatically fail over between zones without interruption. Availability Zones are more highly available, fault tolerant, and scalable than traditional single or multiple data center infrastructures. Aurora DSQL is designed so that you can take advantage of AWS Regional infrastructure while providing the highest database availability. By default, single-Region clusters in Aurora DSQL have Multi-AZ availability, providing tolerance to major component failures and infrastructure disruptions that might impact access to a full AZ. Multi-Region clusters provide all of the benefits from Multi-AZ resiliency while still providing the strongly consistent database availability, even in cases in which AWS Region is inaccessible to application clients.
For more information about AWS Regions and Availability Zones, see AWS Global
Infrastructure
In addition to the AWS global infrastructure, Aurora DSQL offers several features to help support your data resiliency and backup needs.
Backup and restore
During preview, Aurora DSQL doesn't support backup and restore.
Aurora DSQL plans to support backup and restore with AWS Backup console, so you can perform a full backup and restore for your single-Region and multi-Region clusters. What is AWS Backup.
Replication
By design, Aurora DSQL commits all write transactions to a distributed transaction log and synchronously replicates all committed log data to user storage replicas in three AZs. Multi-Region clusters provide full cross-Region replication capabilities between read and write Regions. A designated witness Region supports transaction log-only writes and doesn't use any storage. Witness Regions don't have an endpoint. This means that witness Regions store only encrypted transaction logs, require no administration or configuration, and aren't accessible by users.
Aurora DSQL transaction logs and user storage are distributed across with all data presented to Aurora DSQL query processors as a single logical volume. Aurora DSQL automatically splits, merges, and replicates data based on database primary key range and access patterns. Aurora DSQL automatically scales read replicas, both up and down, based on read access frequency.
Cluster storage replicas are distributed across a multi-tenant storage fleet. If a component or AZ becomes impaired, Aurora DSQL automatically redirects access to surviving components and asynchronously repairs missing replicas. Once Aurora DSQL fixes the impaired replicas, Aurora DSQL automatically adds them back to the storage quorum and makes them available to your cluster.
High availability
By default, single-Region and multi-Region clusters in Aurora DSQL are active-active, and you don't need to manually provision, configure, or reconfigure any clusters. Aurora DSQL fully automates cluster recovery, which eliminates the need for traditional primary-secondary failover operations. Replication is always synchronous and done in multiple AZs, so there is no risk of data loss due to replication lag or failover to an asynchronous secondary database during failure recovery.
Single-Region clusters provide a Multi-AZ redundant endpoint that automatically enables concurrent access with strong data consistency across three AZs. This means that user storage replicas on any of these three AZs always return the same result to one or more readers and are always available to receive writes. This strong consistency and Multi-AZ resiliency is available across all Regions for Aurora DSQL multi-Region clusters. This means that multi-Region clusters provide two strongly consistent Regional endpoints, so clients can read or write indiscriminately to either Region with zero replication lag on commit. Aurora DSQL doesn't provide a managed global endpoint for multi-Region clusters, but you can use Amazon Route 53 as a substitute.
Aurora DSQL provides 99.99% availability for single-Region clusters and 99.999% for multi-Region clusters.
