# Finding cost savings in Azure compute disks | PointFive Blog

Canonical: https://www.pointfive.co/blog/finops-for-azure-compute-disks-hidden-cost-optimization

Evaluate Azure disk capacity, performance, support constraints, and total cost before changing tiers.

By: Jessica Wachtel

Published: 2026-05-06

[All articles](https://www.pointfive.co/blog) 

Jessica Wachtel  Technical Writer  May 6, 2026  3 min read

Choosing an Azure disk involves capacity, IOPS, throughput, latency, resilience, and application requirements. A more expensive tier is not automatically the right choice, and a cheaper tier is not automatically an equivalent replacement.

## Compare the documented disk types

Microsoft lists Standard HDD, Standard SSD, Premium SSD, Premium SSD v2, and Ultra Disks. Premium SSD v2 separates provisioned capacity, IOPS, and throughput and includes a baseline of 3,000 IOPS and 125 MB/s. Its documented maximum throughput is 2,000 MB/s, subject to configuration limits. Both Premium SSD v2 and Ultra Disks are designed for submillisecond latency. Region, VM, OS-disk, redundancy, and feature support differ by disk type. Consult [Microsoft's disk-type documentation](https://learn.microsoft.com/en-us/azure/virtual-machines/disks-types)  for the configuration being evaluated.

These specifications are not a promise that a particular application will see the same latency on both tiers. Nor does a percentage performance or availability target establish a fixed number of daily seconds of latency spikes. Application measurements and the applicable service terms must be evaluated separately.

## Diagnose the bottleneck

Collect representative disk latency, queue depth, IOPS, throughput, capacity, and VM limits. Include peak periods, batch jobs, backups, and startup behavior. A low average can conceal a critical peak; provisioned performance above an observed average can also be intentional headroom.

Distinguish storage limits from CPU, memory, network, and application bottlenecks. Test a candidate configuration against the workload's acceptance criteria before changing production.

## Compare complete costs

Use a dated quote for the same region, currency, redundancy, capacity, and performance requirements. Include transaction charges, bursting where applicable, snapshots, and migration costs. A capacity-only comparison cannot establish equal performance or total savings.

Avoid treating a portal default as proof that a disk is necessary. Equally, do not assume all stateless services or production workloads can use HDD storage without an effect on startup or application behavior.

## Turn the analysis into a controlled change

PointFive OS helps investigate infrastructure usage and prepare optimization actions. Treat potential savings as estimates until the change is implemented and validated against billing. Infrastructure changes require customer-authorized write access and approval before execution.

Record the original configuration, expected impact, rollout plan, rollback path, and post-change measurements. This creates an auditable basis for deciding whether a disk change improved efficiency.

Technical source checked September 17, 2026. Earlier price examples and unsupported daily-latency calculations have been removed.

---
Source: the public page above. Product screenshots and illustrative interfaces are examples, not live customer data.