AWS Aurora MySQL
Google Cloud SQL for MySQL, a single “failover replica” can be created but cannot be used for reads and adding a failover replica will double the cost. Read replicas can be deployed as well, with each increasing the cost of the deployment linearly. It’s not ideal for write intensive workloads due to synchronous replication. A DR node can be deployed in a separate Google Cloud region, but with limited management and orchestration between them, making the solution not geo-ready for 24/7 business-critical environments.
Local High Availability (HA)
Offers automatic failover but only to a single node and it takes 60 seconds. Also there is downtime with maintenance windows.
Multi-Region and DR
Just simple replication to another region. No failback. No active/active. No connection management, orchestration or awareness.
Scalability
Limited to GCP only, can add read replicas but requires a 3rd party proxy or application changes.
Performance
Based on synchronous replication which slows down writes. No smart-scale. And load balancing requires a different 3rd party proxy or application changes.
Deployment and Application Changes
Easy to deploy. Likely requires application changes if anything more than simple failover to a single read node is needed.
TCO
Linear pricing, very expensive Enterprise 24/7 support.
MySQL Version
MySQL Community Edition 5.6, 5.7, 8.0.
Infrastructure and Platform
GCP only.
Completeness
Comes with very basic components that lack functionality. Only simplest local MySQL HA master/slave implementations work out of the box.
Maturity
August 2016.
Security
Can be used in PCI and HIPAA compliance.
Management and Monitoring
Fantastic GUI makes simple cluster management very easy. Alerts must be configured using centralized tool. OS and DB patching is automatic but will cause downtime.