Dashboard: Compute Checks
Thrifty Compute Benchmark
Thrifty developers eliminate their unused and under-utilized compute instances. This benchmark focuses on finding resources that have not been restarted recently, have old snapshots, have high disk IOPS and have large, unused or inactive disks.
Variables
| Variable | Description | Default |
|---|---|---|
| compute_disk_avg_read_write_ops_high | The number of average read/write ops required for disks to be considered frequently used. This value should be higher than compute_disk_avg_read_write_ops_low. | 500 |
| compute_disk_avg_read_write_ops_low | The number of average read/write ops required for disks to be considered infrequently used. This value should be lower than compute_disk_avg_read_write_ops_high. | 100 |
| compute_disk_max_iops | The maximum IOPS allowed for disks. | 2000 IOPS |
| compute_disk_max_size_gb | The maximum size (GB) allowed for disks. | 100 GB |
| compute_running_vm_age_max_days | The maximum number of days virtual machines are allowed to run. | 90 days |
| compute_snapshot_age_max_days | The maximum number of days snapshots can be retained. | 90 days |
| compute_vm_avg_cpu_utilization_high | The average CPU utilization required for virtual machines to be considered frequently used. This value should be higher than compute_vm_avg_cpu_utilization_low. | 35% |
| compute_vm_avg_cpu_utilization_low | The average CPU utilization required for virtual machines to be considered infrequently used. This value should be lower than compute_vm_avg_cpu_utilization_high. | 20% |
Usage
Install the mod:
mkdir dashboardscd dashboardspowerpipe mod initpowerpipe mod install github.com/turbot/steampipe-mod-azure-thriftyStart the Powerpipe server:
steampipe service startpowerpipe serverOpen http://localhost:9033 in your browser and select Compute Checks dashboard.
You could also snapshot and share results via Turbot Pipes:
powerpipe loginpowerpipe benchmark run azure_thrifty.benchmark.compute --shareBenchmark
This dashboard is automatically generated from the following benchmark:
benchmark.compute