An EC2 workload performs large sequential reads. VolumeThroughputExceededCheck remains 1, and application read wait time is increasing.
The volume is gp2, and measurements show a throughput limit rather than an IOPS limit. The team wants stable performance while paying only for the required increase.
Which improvement is most appropriate?
Confirm the throughput limit, move to gp3, set the required throughput explicitly, and verify that the instance provides enough EBS-optimized bandwidth.
Detailed explanation
gp3 lets the team configure IOPS and throughput independently of capacity, while the instance EBS bandwidth must also cover the measured bottleneck.
gp3 lets the team configure IOPS and throughput independently of capacity, while the instance EBS bandwidth must also cover the measured bottleneck.
CloudFront caches HTTP(S) delivery and does not change EBS volume or instance I/O bandwidth.
CloudFront caches HTTP(S) delivery and does not change EBS volume or instance I/O bandwidth.
io2 targets high IOPS and low latency; it is not an automatic answer to a measured sequential-throughput limit.
io2 targets high IOPS and low latency; it is not an automatic answer to a measured sequential-throughput limit.
Increasing gp2 capacity can change its baseline but may add unnecessary cost and removes the evidence needed to verify the fix.
Increasing gp2 capacity can change its baseline but may add unnecessary cost and removes the evidence needed to verify the fix.
Check official documentation
Compare the following points with the official documentation.
Amazon EBS I/O characteristics、gp3 volume performance、EBS-optimized instanceの公式説明を確認する。
Official references
What to confirm
IOPS上限、スループット上限、インスタンス側帯域を区別し、gp3の独立した性能設定を説明できる。
Key points
- gp3
- VolumeThroughputExceededCheck
- EBS-optimized instance
- IOPS
- Throughput
Notes
- Environment: AWS公式SOA-C03試験ガイドとAWS公式ドキュメントの確認
- Confirm the behavior in the official SOA-C03 guide and Amazon EBS documentation.
Foundation review
Two performance boundaries
The lower limit between the volume's provisioned capability and the instance's EBS bandwidth can bottleneck the workload.
Measure before changing
A metric that identifies throughput pressure supports a targeted change instead of an across-the-board volume upgrade.