AWS & CloudAmazon Aurora là cloud-native relational database engine do AWS xây dựng từ đầu, tương thích MySQL và PostgreSQL. Architecture Aurora khác biệt: storage layer tách khỏi compute layer — storage tự động scale đến 128TB, replicate 6 copies trên 3 AZ (4/6 write quorum, 3/6 read quorum), healing tự động khi có node fail; điều này cho phép failover nhanh (<30s) và read replicas lag thấp (<100ms).
Throughput Aurora MySQL nhanh gấp 5x MySQL gốc, Aurora PostgreSQL nhanh gấp 3x PostgreSQL gốc nhờ shared storage architecture loại bỏ replication log shipping. Aurora Global Database: 1 primary region + tối đa 5 secondary regions, replication lag <1s, phù hợp disaster recovery và global read scaling — failover cross-region trong <1 phút.
Aurora Serverless v2 (2022) scale compute theo fractional ACU (Aurora Capacity Unit) từ 0.5 đến 128 ACU, scale up trong vài giây, phù hợp workload có spike bất thường, dev/test environment, multi-tenant SaaS. Aurora Serverless v2 không scale về 0 (khác v1); Aurora Serverless v1 có thể scale về 0 nhưng cold start lâu.
Aurora Machine Learning tích hợp SageMaker và Comprehend trực tiếp từ SQL. Backtrack feature cho Aurora MySQL cho phép rewind database về thời điểm bất kỳ trong 72h mà không cần restore snapshot.
Amazon Aurora is a cloud-native relational database engine built from scratch by AWS, compatible with MySQL and PostgreSQL. Aurora's architecture is distinctive: the storage layer is decoupled from the compute layer — storage auto-scales to 128TB, replicates 6 copies across 3 AZs (4/6 write quorum, 3/6 read quorum), and self-heals when a node fails; this enables fast failover (<30s) and low read replica lag (<100ms).
Aurora MySQL is up to 5x faster than standard MySQL, and Aurora PostgreSQL is up to 3x faster than standard PostgreSQL, thanks to a shared storage architecture that eliminates replication log shipping. Aurora Global Database: 1 primary region plus up to 5 secondary regions, <1s replication lag, suitable for disaster recovery and global read scaling — cross-region failover in under 1 minute.
Aurora Serverless v2 (2022) scales compute in fractional ACUs (Aurora Capacity Units) from 0.5 to 128 ACU, scaling up within seconds — ideal for workloads with unpredictable spikes, dev/test environments, and multi-tenant SaaS. Aurora Serverless v2 does not scale to zero (unlike v1); Aurora Serverless v1 can scale to zero but has a slow cold start.
Aurora Machine Learning integrates SageMaker and Comprehend directly from SQL. The Backtrack feature for Aurora MySQL allows rewinding the database to any point within 72 hours without restoring a snapshot.