Which of the following designs will meet the performance goal MOST cost effectively?
Change the type of Amazon EBS volume from gp2 to io1 and set provisioned IOPS to 9,000.
Increase the size of the gp2 volumes in each instance to 3 TB.
Create a new Amazon EFS file system and move all the data to this new file system. Mount this file system to all 10 instances.
Create a new Amazon S3 bucket and move all the data to this new bucket. Allow each instance to access this S3 bucket and use it for storage.
Explanations:
While changing to io1 with 9,000 provisioned IOPS would meet the IOPS requirement, it is not the most cost-effective solution. Provisioned IOPS (io1) volumes are typically more expensive than increasing the size of gp2 volumes, especially given the current workload’s requirements.
Increasing the size of each gp2 volume to 3 TB will provide an additional 3,000 IOPS per instance (gp2 provides 3 IOPS per GB, so 3 TB = 9,000 IOPS), meeting the performance goal effectively and at a lower cost compared to switching to io1.
Using Amazon EFS might improve performance due to its scalability and throughput capabilities; however, it is generally more expensive than EBS volumes. Moreover, EFS is not optimized for I/O intensive workloads compared to EBS.
Moving data to S3 does not directly solve the IOPS limitation, as S3 is object storage and is not designed for low-latency, high-throughput access typical of I/O intensive workloads. This option would likely increase latency further rather than address the IOPS constraint.
I plan that the answer is:
Increase the size of the gp2 volumes in each instance to 3 TB.