Question #1058
A company operates a high-traffic e-commerce platform using an Amazon RDS for MySQL database with a table containing over 15 million rows. The database utilizes 3 TB of General Purpose SSD storage. The platform processes millions of daily update operations, but the company has observed that some update queries are experiencing delays of 15 seconds or more. Analysis indicates that the storage performance is the primary bottleneck.
Which solution will resolve this performance issue?
Migrate the storage type to Provisioned IOPS SSD.
Switch the DB instance to a compute optimized instance class.
Convert the DB instance to a burstable performance instance class.
Implement Multi-AZ RDS read replicas using MySQL semi-synchronous replication.
Explanation
The issue arises from storage performance limitations in a high-transaction RDS MySQL environment. General Purpose SSD (gp2) offers baseline IOPS (3 IOPS/GB) and burst credits. At 3 TB, the baseline is 9,000 IOPS. Sustained high I/O workloads exhaust burst credits, causing latency.
Why A is correct: Provisioned IOPS SSD (io1/io2) allows provisioning dedicated IOPS (e.g., 20,000+), ensuring consistent performance for write-heavy workloads like millions of daily updates.
Why others are incorrect:
- B: Compute-optimized instances address CPU/memory bottlenecks, not storage I/O.
- C: Burstable instances (e.g., t-class) rely on CPU credits and are unsuitable for sustained high I/O.
- D: Read replicas scale read traffic, not write performance. Multi-AZ improves availability but doesn’t resolve storage I/O bottlenecks.
Key Points:
- Use Provisioned IOPS SSD for I/O-intensive workloads requiring predictable performance.
- General Purpose SSD is cost-effective but limited by burst credits for sustained high I/O.
- Storage, not compute or replication, is the bottleneck here.
Answer
The correct answer is: A