Skip to content

Latest commit

 

History

History
83 lines (62 loc) · 4.91 KB

README.md

File metadata and controls

83 lines (62 loc) · 4.91 KB
Raw

ebs-volume-usage-exporter

On average, 70% of cloud block storage, such as EBS, is under-utilized. That typically means that many storage consumers (e.g., databases) request more storage than they need, creating overall storage waste. Simplyblock’s EBS Volume Usage Calculator helps you understand the usage of persistent storage inside an Amazon EKS cluster and identify opportunities for EBS cost optimization. This isn’t always easy since most volumes are dynamically provisioned, so there is no general overview of how much storage is being actually used.

The exported CSV can be analyzed manually or automatically using our EBS Volume Usage Calculator. The calculator runs completely local (in your browser) and does not upload information to our servers.

What is collected?

The ebs-volume-usage-exporter collects some basic volume data (e.g., aws region, provisioned capacity, IOPS, throughput, name, ...), as well as usage information (avg / max IOPS and throughput).

The following list has all exported CSV columns and their meaning:

  • aws_region: The volume's AWS region (for multi-region clusters)
  • pv_name: The volume's UUID-based volume name
  • pv_size: The volume's human-readable provisioned capacity
  • ebs_volume_id: The volume's Amazon EBS volume id
  • ebs_volume_type: The volume's Amazon EBS volume type (one of sc1, st1, gp2, gp3, io1, io2)
  • ebs_size_gb: The volume's provisioned capacity (as integer)
  • ebs_provisioned_iops: The volume's provisioned IOPS (can be "N/A" in case of default)
  • ebs_provisioned_throughput: The volume's provisioned throughput (can be "N/A" in case of default)
  • read_io_avg: The volume's average read IOPS
  • read_io_max: The volume's maximum read IOPS
  • write_io_avg: The volume's average write IOPS
  • write_io_max: The volume's maximum write IOPS
  • read_mbps_avg: The volume's average read throughput
  • read_mbps_max: The volume's maximum read throughput
  • write_mbps_avg: The volume's average write throughput
  • write_mbps_max: The volume's maximum write throughput
  • snapshots: The volume's snapshot as a comma-separated list
  • gigabytes_available: The volume's free disk space
  • start_time: The start time of the measurement period
  • end_time: The end time of the measurement period

A CSV file content may look like this:

aws_region,pv_name,pv_size,ebs_volume_id,ebs_volume_type,ebs_size_gb,ebs_provisioned_iops,ebs_provisioned_throughput,read_io_avg,read_io_max,write_io_avg,write_io_max,read_mbps_avg,read_mbps_max,write_mbps_avg,write_mbps_max,snapshots,gigabytes_available,start_time,end_time
us-east-2,pvc-5a6f744a-2bb0-4b89-889f-bb66bcbd1f63,150Gi,vol-0e41cb8b3856d5d12,st1,150,N/A,N/A,0.86,34.2,71.61,1164.2,0,0,0.06,0.94,,146.57,2024-12-08T14:12:48.590302+00:00,2024-12-09T14:12:48.590302+00:00

Collector Usage

The collector is provided as a ready-to-use helm chart which uses the default python image to execute the collector script.

To deploy the helm chart, a few values need to be configured in the values.yaml file. Therefore, the following information needs to be available:

  • Amazon EKS cluster with cluster-admin privilege
  • S3 bucket

First, we need to create the necessary secret. To create it you need a AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY. The following permissions are required:

  • IAM EC2: Read-Only
  • Cloudwatch: Read-Only
  • S3 Bucket: Full

Use the following command to create the secret:

kubectl create secret generic aws-secret \
  --from-literal=AWS_ACCESS_KEY_ID=your-access-key-id \
  --from-literal=AWS_SECRET_ACCESS_KEY=your-secret-access-key

Afterward, fill in the name of the secret (in this case aws-secret) into the values.yaml under AWS_SECRET_REF.

Secondly, we need to create a service account and cluster role (get and list PVs, PVCs, and snapshots). You can do this with the following commands:

kubectl create sa metric-aggregation-sa
kubectl apply -f manifests/storage-cluster-role.yaml
kubectl apply -f manifests/storage-cluster-role-binding.yaml

Finally, install the helm chart and let the tool run. After the successful export, you'll find the exported CSV file in the S3 bucket you provided. To kick off the installation, use the following command. The TIME_DURATION parameter defines the number of days for the data collection.

helm install metric-agg charts/pv-metrics-aggregation/ \
  --set scriptConfig.S3_BUCKET_NAME=metrics-aggragation \
  --set scriptConfig.AWS_REGION=us-east-1 \
  --set scriptConfig.TIME_DURATION=1 \
  --set scriptConfig.CLUSTER_NAME=random

If you want, you can now upload your exported CSV file to our EBS Volume Usage Calculator for automatic analysis and calculation of your overprovisioning. The calculator runs completely local (in your browser) and does not upload information to our servers.