Skip to content

DataWriteIOBytes jump after switching EFS Throughput mode from bursting to elastic

0

I have EC2 instance, running Amazon Linux 2 with mounted EFS disk. There is a java application running on the instance and reading/writing files to the efs disk. I have the need to test switching the Throughput mode from bursting to elastic and right after that, I noticed jump on DataWriteIOBytes. Normally, DataWriteIOBytes is constant at 5MB per minute and DataReadIOBytes is 0 most of the time. After the switch to elastic, everything seems good but these usual 5mb DataWriteIOBytes per minute now changed to about 39MB. There were no changes to the application, no even restart, it didn't change its operations in any way. The jump on the DataWriteIOBytes matches with the jump on the PermittedThroughput, so it seems to be that the change of the throughput mode somehow changed the DataWriteIOBytes measurement, but why?

1 Answer
0

The jump in DataWriteIOBytes after switching from Bursting to Elastic throughput mode is likely related to how Amazon EFS measures and reports throughput metrics rather than an actual change in your application's behavior.

When you switch to Elastic throughput mode, the way throughput is calculated and reported changes. In Elastic throughput mode, you're billed for the amount of data transferred (reads and writes), which means the system may be more precisely measuring and reporting the actual throughput being used.

The PermittedThroughput metric represents the maximum allowed throughput for your file system, which is calculated differently based on the throughput mode:

  • For Bursting throughput, it's a function of file system size and BurstCreditBalance
  • For Elastic throughput, it reflects the maximum write throughput of the file system

When you switched to Elastic throughput, the PermittedThroughput value likely increased, which corresponds with the jump you observed in DataWriteIOBytes. This doesn't necessarily mean your application is writing more data - it's more likely that the measurement and reporting of the throughput has changed.

The CloudWatch metrics for throughput monitoring (including WriteIOBytes) represent the actual throughput on your file system, but the way these metrics are calculated and reported can vary based on the throughput mode. Since your application didn't change its operations, the most probable explanation is that the switch to Elastic throughput mode altered how the write operations are being measured and reported rather than changing the actual amount of data being written.
Sources
Using CloudWatch metrics for Amazon EFS - Amazon Elastic File System
Managing file system throughput - Amazon Elastic File System
Amazon Elastic File System (EFS) | Cloud File Storage | FAQs

answered a year ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.