Globally-Distributed Databases and Going Multi-Regional for Optimized Data Recovery

Enterprise growth is a goal for most companies, yet it concurrently requires strategies in the event of data disaster or to maintain a regularly operable, productive system. With a cloud-native managed, multi-region infrastructure, enterprises can use their global presence as an advantage towards recovery methods.

DBTA held a webinar, “Multi-Active Cloud-Native Database,” featuring speaker Chad Tindel, principal NoSQL solutions architect for Amazon Web Services, to discuss DynamoDB’s multi-region, disaster-prevention capabilities alongside Amazon DynamoDB Global Tables.

DynamoDB’s capacity for multi-region server management provides a replication-based cushion in the event of disaster recovery scenarios, as well as for business continuity’s sake.

If you deploy applications in multiple regions within AWS, each of those applications will have an assortment of microservices with an additional underlying AWS service—presenting a prime opportunity for business failure if code stops working correctly, or errors are sustained consistently from cloud services. With multi-region application support, if a server fails, the architecture can be migrated to a different region for uninterrupted operability. A highly-distributed geographical customer base would also be a great candidate for multi-region architecture, according to Tindel, considering its ability to lower latency to overall improve the end-user experience. 

Naturally, DynamoDB still functions well with single region systems. Yet it cannot provide consistent low latency, in a global sense, or 99.999% availability SLA (compared to a single region’s availability SLA of 99.99%).

Backup and restore options with DynamoDB, as a result of multi-region architectures, enables several modes of recovery that can revive systems in case of failure. With on-demand backups, users can acquire a “moment in time snapshot” of the data for long-term, data archival and compliance. Point-in-time-restore, or PITR, allows for short term retention and data corruption protection through a holistic, continuously running log of transactions from the last 35 days. Regardless of your choice in backup and recovery, DynamoDB allows for hundreds of TB to be backed up instantly with no performance impact to your data tables.

Amazon DynamoDB’s Global Tables are comprehensive tools to ensure recovery within a replication relationship across servers, around the world. With two separate DynamoDB tables in each region that have an established replication relationship, data is synched cross-regionally and bi-directionally. This provides the ability to write and read in all tables simultaneously and globally. Writes to any replicas are repeated in all other tables, with more than 27 regions available now.

Tindel stressed viewers to understand their enterprise missions more deeply to determine if a multi-region architecture would suit their needs. Considering how much data you can afford to recreate or lose, how quickly you must recover systems in event of a failure, and what the cost of downtime is, are critical in evaluating what your enterprise requires for recovery and backup instances.

To see examples and demos of DynamoDB’s multi-region capabilities from Tindel, as well as for more information about global-scale operations strategies, you can view an archived version of the webinar here