Amazon Aurora DSQL (Data column)
Werner Vogel, CTO Amazon, in his keynote at re:Invent 2024 Half an hour told about the new capabilities of the Aurora base, which appeared about 10 years ago, and now moved to the class newSQL and began to resemble the features of Google Spanner, which just appeared more 10 years ago. Here's what Werner did. 1) Serverless Architecture - for automatic scaling under the desired loads 2) High Availability - for one region 99.99% availability for multi-regional master-master setup 99.999% 3) PostgreSQL Compatibility Full PostgreSQL support 4) Performance They promise 4x speed compared to other similar solutions, apparently from the world of newSQL. I would look at the benchmarks to see who they were compared to:) 5) Active-Active Multi-Region Operations A multi-regional version that can take records and replicate them in real time. (It's interesting to see how fast this works in multi-regional performance.) 6) Zero Infrastructure Management Users should not think about info. (Except how much that happiness will cost.) 7) Innovative Transaction Processing - here Werner pathetically telling about the analogue of the Google TrueTime, which was still 10+ years ago at Google Spanner. Amazon now has its own satellites and atomic clocks for microsecond accuracy, all called the Amazon Time Sync Service. This allows you to make distributed transactions with good consistency guarantees:)
P.S. I’ve mentioned Aurora a lot before.
- Amazon Aurora: Design Considerations for High Troughput Cloud-Native Relational Databases - AWS re:Invent 2023 - [LAUNCH] Achieving scale with Amazon Aurora Limitless Database (DAT344)
#Databases #Architecture #SoftwareArchitecture #SystemDesign #DistributedSystems #Cloud #Management #Leadership