How to change volumes specs on RTS for better performance?

Question

After a series of tests we have found that volume specs is what most impacts the RTS performance.

Here are the test details:
Changing the volume to type=gp3 / IOPS=16000 / Throughput = 1000 (screenshot below) and running on r6i.16xlarge PostgreSQL EC2
- Test case runs in 9 minutes. 
- Same test case runs in 50 minutes on the same EC2 and using default gp2 volume spec.

Now running the standard PostgreSQL on t3.2xlarge with the same high performance volume (type=gp3 / IOPS=16000 / Throughput = 1000) 
- Test case runs in 12 ½ minutes.

Jean N · Accepted Answer

Hi Frédéric,

The previous answer is correct. Additionally, for RTS, you can directly modify the EBS Snapshots. To do this:

1) Edit the tags of the snapshots
    2) Update the following tags:
        - dws:snapshot:disk:io: Set this to your desired number of IOps (maximum 16,000 for gp3)
        - dws:snapshot:disk:type: Set this to 'gp3'

After making these changes, all new volumes created from these snapshots will automatically inherit the specifications set in the tags. This ensures that your new volumes will have the desired gp3 type and IOps settings.

gtrelles · Answer

Hi Frederic,

To change the volume specs go to Amazon EC2 Console and select Volumes, and pick the volume that needs to be modified. You will be able to modify the volume type, size, IOPS and throughput.

This process does not restart your instance nor does it detach the volume, which means there is no interruption whatsoever to your application.

Depending on the volume size and usage, it might take some time to change to the new volume type.

re:Post Agent · Answer

To improve RDS (Amazon Relational Database Service) performance through volume specifications, you can consider the following steps:

1. Migrate to gp3 volumes: Amazon EBS gp3 volumes offer higher baseline performance than gp2 volumes, providing 3,000 IOPS and 125 MiB/s throughput regardless of volume size. This can significantly improve performance at a potentially lower cost.

2. Increase IOPS: For gp3 volumes, you can provision up to 16,000 IOPS. Your test results show a significant performance improvement when using 16,000 IOPS, reducing the test case runtime from 50 minutes to 9 minutes.

3. Increase throughput: gp3 volumes allow you to increase throughput up to 1,000 MiB/s. Your test case benefited from setting the throughput to 1,000 MiB/s.

4. Choose appropriate instance types: While volume specs play a crucial role, instance type also matters. Your tests show that an r6i.16xlarge instance performed better than a t3.2xlarge instance, even with the same high-performance volume settings.

5. Monitor and adjust: Use Amazon CloudWatch to monitor metrics like DiskQueueDepth. If you notice high values, it may indicate that your storage is not keeping up with I/O requests.

6. Consider Provisioned IOPS (io1/io2) for extremely high-performance needs: If gp3 doesn't meet your performance requirements, you might want to explore io1 or io2 volumes, which can provide even higher IOPS.

Remember that changing volume types (e.g., from gp2 to gp3) can be done through the AWS Management Console or AWS CLI. However, be cautious when increasing volume size beyond 16 TiB with io2 volumes, as you won't be able to revert to gp2, gp3, or io1 without creating a new instance and migrating your database.

Always test thoroughly before making changes to your production environment, as performance can vary based on specific workloads and configurations.  
**Sources**  
[ModifyVolume - Amazon Elastic Compute Cloud](https://docs.aws.amazon.com/AWSEC2/latest/APIReference/API_ModifyVolume.html)  
[Selecting IOPS and throughput when migrating to gp3 Amazon EBS volume types - Amazon EMR](https://docs.aws.amazon.com/emr/latest/ManagementGuide/emr-plan-storage-gp3-migration-selection.html)  
[Best practices for Amazon RDS for SQL Server with Amazon EBS io2 Block Express volumes up to 64 TiB | AWS Database Blog](https://aws.amazon.com/blogs/database/best-practices-for-amazon-rds-for-sql-server-with-amazon-ebs-io2-block-express-volumes-up-to-64-tib/)  
[Maximizing Storage Performance and Savings with Amazon EBS gp3 and Ollion | AWS Partner Network (APN) Blog](https://aws.amazon.com/blogs/apn/maximizing-storage-performance-and-savings-with-amazon-ebs-gp3-and-ollion/)  
[Troubleshoot DiskQueueDepth increase in RDS for PostgreSQL | AWS re:Post](https://repost.aws/knowledge-center/rds-postgresql-diskqueuedepth)

How to change volumes specs on RTS for better performance?

Add your answer

Relevant content