# Azure Rightsizing: VMs, Managed Disks, SQL, Cosmos DB, Logs

> Azure rightsizing in ZopNight: VM downscales, Premium disk tier and IOPS steps, SQL DTU tiers, elastic pools, Cosmos DB RU/s, storage and log retention.

Source: https://zop.dev/recommendations/azure/rightsizing
Published: 2026-07-01 · Updated: 2026-09-24 · Author: avinash-gaurav · Tags: zopnight, recommendations, rightsizing, azure

---

ZopNight treats Azure sizes and tiers as ladders, and most Azure rightsizing findings are a one-rung step down, priced from both rungs.

ZopNight's Azure rightsizing rules model VM sizes, disk performance tiers, SQL service objectives, App Service Plan tiers and storage redundancy levels as ladders of SKUs. They measure where a resource actually sits and propose the next rung down, pricing the move from Azure's retail rates for both the current and target SKU. Where no safe lower rung exists, the rule abstains.

## Virtual machines

**Azure VM Rightsizing, Downscale Opportunity** names a smaller VM size and computes the saving from the real on-demand rate difference. It is one of the rules wired to remediation as an in-place resize, with the usual approval step for non-admins. See <a href="https://zop.dev/integrations/azure/vm">Azure VMs</a> and <a href="https://zop.dev/solutions/azure-virtual-machines">Azure Virtual Machines</a>.

## Managed disks, three ways down

Three rules split Premium SSD disks between them so they never double-count:

- **Azure Small Premium Disk Tier Optimization** owns small Premium disks provisioned under 500 IOPS on production or unclassified workloads, and suggests Standard SSD.
- **Azure Non-Prod Premium Disk Downgrade** acts only on disks with positive non-production evidence, and never touches disks named for database, log or tempdb roles.
- **Azure Premium Disk IOPS Downgrade** walks a larger Premium disk down its P-tiers, and needs measured read and write operations before it will fire.

**Azure VM Ultra Disk with Low IOPS** covers Ultra disks. All three Premium rules can run as resizes through the remediation wizard. Fact sheet: <a href="https://zop.dev/integrations/azure/disk">managed disks</a>.

## Databases

- **Azure SQL Right-sizing** steps a DTU database down one service objective, for example S3 to S2, when DTU usage is low over 30 days.
- **Azure SQL Multiple Small DBs Could Use Elastic Pool** compares the standalone cost of several low-DTU databases on one server with the cost of a pool.
- **Azure Cosmos DB Provisioned Throughput Rightsizing** sizes a new RU/s value from the measured peak plus burst headroom, on manually provisioned throughput.

See <a href="https://zop.dev/integrations/azure/sqldatabase">Azure SQL Database</a>, <a href="https://zop.dev/integrations/azure/cosmosdb">Cosmos DB</a> and <a href="https://zop.dev/solutions/azure-sql-database">Azure SQL Database</a>.

## Storage, logs and backup

- **Azure Dev/Test Storage Using GRS Replication** prices the geo-replication premium a dev/test account pays over LRS.
- **Azure Storage Account Hot Tier with Low Access** prices the Hot to Cool difference on the account's measured size.
- **Azure Log Analytics Excessive Retention** fires above 90 days of interactive retention and prices the saving from the workspace's actual billed retention charge.
- **Azure Recovery Vault Geo-Redundant Downgrade** and **Azure Recovery Vault Retention Tune** cover backup storage.

## Plans and pools

**Azure App Service Plan Right-sizing** steps a plan one tier down and is wired to remediation. **Azure Function Premium Plan with Low Executions** and **AKS Node Pool Underutilized** cover serverless and Kubernetes capacity; see <a href="https://zop.dev/solutions/azure-aks">Azure AKS</a>.

Every Azure rule is listed in <a href="https://zop.dev/integrations/azure/recommendations">Azure recommendations</a>. Cross-cloud view: <a href="https://zop.dev/recommendations/rightsizing">rightsizing</a>.