Supported Databases › MySQL Aurora
MySQL Aurora Performance Investigation with MCP Workflows
Amazon Aurora MySQL combines MySQL compatibility with Aurora's distributed storage architecture. The storage layer eliminates some traditional MySQL replication bottlenecks but introduces Aurora-specific performance characteristics that require their own investigation approach. DatabaseMCP covers both MySQL-level diagnostics and Aurora-layer metrics in a unified workflow.
Common MySQL Aurora performance problems
- AuroraReplicaLag growth: Unlike standard MySQL replication, Aurora reader nodes share the same storage volume as the writer. Replica lag should typically be very low, but heavy DDL operations, large transactions, or network pressure between nodes can cause AuroraReplicaLag to grow and serve stale reads.
- Post-failover query plan instability: When Aurora promotes a reader to writer following a failover, the new writer starts with a cold buffer pool. Queries that relied on cached pages from the previous writer experience higher I/O until the buffer pool warms up.
- Aurora Serverless v2 capacity ceiling: Serverless v2 scales ACU capacity automatically but has a defined maximum. Workload spikes approaching the max ACU limit cause latency increases before scaling catches up, or fail to scale if the max is set too conservatively.
- Connection routing errors: Aurora provides separate writer and reader endpoints. Applications connecting write traffic to a reader endpoint or connecting all traffic to the writer unnecessarily degrade performance and eliminate horizontal read scaling.
- VolumeWriteIOPs saturation: Aurora's distributed storage is fast but write-intensive bulk operations, ETL jobs, or large transaction commits can saturate the write path and affect overall instance throughput.
- Core MySQL issues: InnoDB lock contention, slow queries, buffer pool pressure, and connection saturation all apply to Aurora MySQL the same as any MySQL deployment.
MCP investigation workflow for MySQL Aurora
- Aurora CloudWatch metrics: AuroraReplicaLag, ServerlessDatabaseCapacity, VolumeWriteIOPs, BufferCacheHitRatio, and DatabaseConnections give a fast view of Aurora-layer health before query-level investigation.
- Performance Insights: Aurora Performance Insights shows database load broken down by wait events, SQL statements, and hosts. It is the primary tool for identifying query-level load drivers on Aurora MySQL.
- Performance Schema query digest: Identify top query patterns by total latency and execution count. Review rows examined vs rows sent for index coverage gaps.
- Failover and scaling event correlation: Review Aurora Events for failover and scaling history. Match latency spikes against these events to separate infrastructure causes from query or schema issues.
- InnoDB and connection health: Review buffer pool hit rate, lock waits, deadlock frequency, and current connection distribution across writer and reader nodes.
Related MySQL and Aurora coverage
For standard RDS MySQL deployments see MySQL RDS investigation. For self-hosted MySQL see MySQL investigation. For Aurora PostgreSQL see PostgreSQL Aurora investigation.
Get started
See platform features on minidba.com