Overview
Lucidity scaling policies can include optional IOPS and throughput targets. When set, Lucidity enforces these values on every disk that the autoscaler creates or resizes, so your performance floor is maintained automatically, without manual intervention after each scale event.
This article explains how performance settings work, which disk types are supported, and what to check before assigning a policy that includes IOPS or throughput targets.
Supported Disk Types
IOPS and throughput enforcement applies only to the following disk types:
AWS: gp3 (EBS)
Azure: Premium SSD v2
GCP: Hyperdisk Balanced, Hyperdisk Extreme
If your instance runs on a different disk type (for example, AWS gp2, Azure Premium SSD v1, or Azure Standard SSD), assigning a policy with IOPS targets will not cause errors; the performance settings will not be enforced. Scaling will follow standard utilization rules only.
How Enforcement Works
After every disk create or resize event triggered by the autoscaler, Lucidity applies the IOPS and throughput values from the policy to the resulting disk. There is no manual step required after a scale event.
IOPS cap based on disk size
Each cloud provider sets a maximum IOPS that a disk can support based on its size. If the policy target exceeds what the current disk size allows, Lucidity applies the maximum the disk supports, not the policy value. The gap is logged, and a warning is shown in the VM detail view. Once the disk grows to a size that supports the full policy target, enforcement catches up automatically.
Update cooldown
Lucidity enforces a 24-hour cooldown per disk between IOPS and throughput updates. This applies across all cloud providers. If a disk has already received an IOPS or throughput update within the past 24 hours, Lucidity queues the next update rather than skipping it. The disk resize completes immediately; the IOPS update follows once the cooldown clears.
This cooldown is intentionally more conservative than the individual cloud provider limits. Lucidity may have concurrent expand or shrink operations in flight on the same disk pool; allowing multiple IOPS updates during active scaling operations could cause system inconsistencies. The 24-hour window per disk eliminates that risk.
Cloud-Specific Constraints
AWS gp3
IOPS range: 3,000 to 16,000
Maximum IOPS = disk size in GB x 500
Throughput range: 125 to 1,000 MB/s
Maximum throughput = provisioned IOPS x 0.25
Required IAM permission:
ec2:ModifyVolume
Azure Premium SSD v2
Baseline (disks up to 6 GiB): 3,000 IOPS / 125 MB/s
For disks above 6 GiB: IOPS cap increases by 500 IOPS per GiB
Throughput above 1,000 MB/s requires IOPS of at least 5,000
GCP Hyperdisk Balanced
IOPS range: 3,000 to 160,000
For disks under 320 GiB: maximum IOPS = min(500 x size in GiB, 160,000)
Throughput range: 140 to 2,400 MiB/s
First 3,000 IOPS and 140 MiB/s are included in the base price; only provisioning above these values is billed additionally
GCP Hyperdisk Extreme
IOPS is configurable
Throughput cannot be set independently; it is derived from IOPS
Impact on Savings
Provisioned IOPS and throughput above each cloud provider's baseline are billed separately. Lucidity's savings calculations reflect this: for instances with a performance policy active, the savings figure shown is the net of storage savings minus IOPS and throughput overhead.
If IOPS overhead fully offsets storage savings, the savings cell shows $0. Clicking the cell shows a breakdown: storage savings, IOPS cost, and net figure.
CSP baselines used in the calculation (no additional charge below these values):
Cloud / Disk Type | Baseline IOPS | Baseline Throughput |
|---|---|---|
AWS gp3 | 3,000 | 125 MB/s |
Azure Premium SSD v2 | 3,000 | 125 MB/s |
GCP Hyperdisk Balanced | 3,000 | 140 MiB/s |
Before You Assign a Performance Policy on AWS
Applying IOPS settings on AWS gp3 requires the ec2:ModifyVolume permission in your IAM role. If this permission is missing:
The policy can still be saved and assigned
Disk resizes will succeed as normal
IOPS updates will fail at scale event time, with a permission-specific error shown in the Jobs view
To avoid enforcement gaps, confirm ec2:ModifyVolume is present in your AWS account's IAM role before assigning a policy with IOPS targets.
What Happens When You Clear IOPS Settings from a Policy
Removing IOPS and throughput values from an existing policy does not reset already-provisioned disks. Existing disks retain their current provisioned values. New disks created after the change receive cloud default settings.