SQL - Aurora

Question 1

What is Aurora?

Answer 1

o Aurora is a DB engine developed by AWS that is compatible with MySQL and PostreSQL and associated tools
o Aurora has a radically different architecture a opposed to the other RDS database engines:
 Uses a base configuration of a cluster
 A cluster contains a single primary instance and zero or more replicas (up to 15), with synchronous replication

Question 2

What are the characteristics of Aurora’s Cluster Storage?

Answer 2

 All instances (primary and replicas) use the same shared storage – the cluster volumes
 Cluster Volume is totally SSD based, and can scale up to 64 TiB in size
 Replicates data 6 times, across 3 AZs
 Aurora can tolerate 2 failures with no write impacted, and 3 failures with no read impacted
 Aurora storage is auto healing

Question 3

How does Cluster Scaling and Availability work in Aurora?

Answer 3

 Cluster Volumes scale automatically, are billed only for consumed data, and are constantly backed up to S3 (high watermark approach – if you go up to 10Tb, and then delete down to 5Tb, you are still charged for 5Tb, unless you delete the DB and rebuild it from the 5Tb instance)
 Aurora replicas improve availability, can be promoted to be a primary instance quickly, and allow for efficient read scaling
 Each replica is given a specific Tier (0-15), and failover happens following the Tier levels (0 -> 1 -> 2 -> …)
 Reads and Writes use the cluster endpoint
 Reads can use the reader endpoint, which balances connections over all replica instances

Question 4

What is Aurora’s Backtracking capability?

Answer 4

 Aurora supports backtracking, which allows restoring the DB to any point in the past, up to 72 hours
 During the backtracking process, the whole cluster is out of service, but you have the flexibility of not needing to restore to a new DB (as with RDS)

Question 5

What are Aurora’s Parallel Query and Aurora Global?

Answer 5

o Parallel Query is an Aurora feature, that can be enabled at DB creation, that improves the performance of analytic queries by pushing processing down to the Aurora storage layer (all nodes executing the query at the same time). This is especially useful for hybrid transactional/analytic workloads
o Aurora Global (as opposed to Regional) can be selected at DB creation time, and enables provisioning of the Aurora DB in multiple AWS Regions. Writes in the primary AWS Region are replicated with typical latency of <1sec to secondary AWS Regions

Question 6

What is Aurora Serverless?

Answer 6

o Aurora Serverless (as opposed to Provisioned) is based on the same DB engine as Aurora, but instead of provisioning certain resource allocation, Aurora Serverless handles this as a service; it is based in a single AZ, it is not accessible via a VPN or an inter-region VPC peer
o You need to specify the minimum and maximum number of Aurora Capacity Units (ACUs)
o Aurora Serverless can use the Data API, which is needed for example for Query Editor
o Employs the concept of a Proxy Fleet, which owns the mapping of requests from applications to the Aurora Serverless Cluster
o Removes most of the complexity in managing DBs and Capacity
o Charges are based on the DB resources used, on a per-second basis; it supports pausing, allowing charges for storage only if DB usage drops to zero
o Ideal use cases: deploying a new application and unsure of usage patterns; applications with strong peaks; developers using DBs during work hours only