Azure NetApp Files Service Levels: Standard to Flexible

Introduction

Azure NetApp Files (ANF) teams hit the same wall again and again. A small transactional database needs serious throughput, but only a sliver of capacity. A large file share needs terabytes, but barely any performance.

Fixed service levels tie the two together, so you end up paying for one to get the other. ANF now offers Standard, Premium, Ultra, and Flexible levels to address this, each balancing performance, capacity, predictability, and cost differently.

This guide gives you a decision framework: how each level works, how Flexible separates capacity from throughput, and which workload traits should drive your choice. Verify every limit and price against Microsoft documentation before you deploy.

Key Takeaways

  • Standard, Premium, and Ultra tie throughput to capacity; Flexible sizes them independently.
  • Choose by throughput, capacity, latency, variability, protocol, and recovery needs, not capacity alone.
  • Flexible cuts overprovisioning only when capacity and performance don't grow together.
  • Validate pricing, quotas, and change rules against current Microsoft documentation.

Azure NetApp Files Service Levels: Standard, Premium, Ultra, and Flexible

ANF uses a simple hierarchy: a regional NetApp account contains capacity pools, and volumes live inside those pools. The service level is a property of the capacity pool, with one level per pool.

Billing centers on the pool, meaning provisioned capacity (plus throughput on Flexible), not simply the data you store.

Throughput depends on the QoS type. With automatic QoS, volume quota plus service level sets throughput. With manual QoS, you assign throughput to each volume.

Standard

Standard delivers up to 16 MiB/s per provisioned TiB. It suits general-purpose file shares and workloads with modest performance needs. Its predictable profile can beat Flexible when capacity and demand grow together and you want simple budgeting.

Regular volumes cap at 1,600 MiB/s on Standard.

Premium

Premium provides up to 64 MiB/s per TiB, with a regular-volume ceiling of 4,500 MiB/s. It fits business applications, file services, VDI, and moderately demanding databases.

Premium bundles capacity and performance, while Flexible lets you set each independently.

Ultra

Ultra delivers up to 128 MiB/s per TiB and also tops out at 4,500 MiB/s on regular volumes. It targets the most demanding workloads: high-transaction databases, SAP or Oracle, HPC, and intensive analytics.

One caution: Microsoft says large volumes are not suited to SAP HANA, Oracle, SQL Server, or other database data and log volumes. Use regular volumes there, with application volume groups where appropriate.

Flexible

Flexible, generally available since September 2025, provisions capacity and throughput independently. The key mechanics:

  • 128 MiB/s baseline is included per pool, regardless of pool size
  • Additional throughput is added in 1 MiB/s increments without adding capacity
  • Maximum pool throughput is 5 x 128 MiB/s per TiB of pool size, so a 1-TiB pool ranges from 128 to 640 MiB/s
  • Requires new manual-QoS pools
  • Throughput increases anytime; decreases require a 24-hour cooldown after any pool-throughput change

Flexible Azure NetApp Files throughput and capacity model diagram

Per-volume limits still apply.

How Flexible Changes Azure NetApp Files Capacity and Throughput Economics

The Overprovisioning Problem

Picture a 1-TiB analytics dataset that needs 512 MiB/s. On a fixed tier you'd buy 4 TiB of Ultra just to reach that throughput, paying for 3 TiB you never use. Flip it around: a 175-TiB DR replica or archive share needs almost no performance, yet you pay a performance-bearing rate for it.

Flexible changes the question from "which tier gives me enough performance?" to "how much capacity and how much throughput does this workload need?"

Pricing Mechanics

Using the East US list prices on Microsoft's pricing page (USD, single encryption, observed November 2026; verify current rates):

Meter Price per month
Standard capacity $0.14746/GiB
Premium capacity $0.29419/GiB
Ultra capacity $0.39274/GiB
Flexible capacity $0.11023/GiB
Flexible throughput above baseline $2.24986 per MiB/s

Flexible capacity costs less than every fixed tier, and the first 128 MiB/s per pool is free. You pay extra only for throughput beyond that.

Published Cost Examples

Microsoft's cost model documentation offers illustrations at representative rates. They are not quotes or measured customer savings.

Scenario Fixed tier Flexible Saving
Analytics, 1 TiB at 512 MiB/s 4 TiB Ultra, $1,609/month $976/month 39%
DR, 175 TiB Standard, $26,425/month $19,753/month 25%
Capacity-heavy, 550 TiB at 128 MiB/s 550 TiB Standard 550 TiB, baseline 25%

Azure NetApp Files fixed tier versus Flexible cost savings comparison

Savings depend on your workload, and these figures hold only under the stated assumptions.

Workloads That Benefit Most

  • High-performance databases with small datasets
  • Oracle or SAP demand that swings over time
  • Dev/test environments built from short-term clones
  • DR replicas and large file shares with moderate throughput

Cost-Modeling Method

  1. Pull peak, average, and minimum throughput for each volume.
  2. Record provisioned and actually needed capacity.
  3. Price the current fixed-tier design.
  4. Price a Flexible design: capacity plus throughput above 128 MiB/s per pool.
  5. Label the difference as workload-dependent.

Trade-offs

  • Under-measuring demand leads to under-provisioned throughput.
  • Flexible pools can't be converted to or from Standard, Premium, or Ultra pools.
  • Regional availability and quotas need checking.
  • You must monitor and rightsize more actively.
  • Throughput decreases are limited by the 24-hour cooldown.

Choosing the Right ANF Service Level for Each Workload

Start with the criteria that matter for each workload:

  • Throughput and IOPS needs
  • Latency sensitivity
  • Capacity growth and variability
  • Read/write profile and protocol
  • Availability and replication design
  • Tolerance for ongoing rightsizing

Standard, Premium, or Ultra for Predictable Demand

Fixed levels win when demand is steady, budgets need to be simple, and capacity and throughput grow together. Fewer knobs means less tuning.

Signal Likely fit
Large, mostly idle shares Standard
Business apps, VDI, file services Premium
Consistently heavy database or HPC I/O Ultra

Confirm against current Microsoft limits, not arbitrary thresholds.

Flexible for Decoupled or Variable Requirements

Flexible fits when you see:

  • High throughput with limited capacity
  • High capacity with moderate throughput
  • Pronounced peaks or temporary bursts
  • Different production and DR performance needs

It isn't automatically cheapest. Compare total cost across capacity, baseline and added throughput, replication, backups, snapshots, and management effort.

Workload Scenarios

  • Transactional databases: Small capacity, high performance. Evaluate Flexible or Ultra on regular volumes.
  • SAP or Oracle: Variable demand. Flexible suits swings; Ultra suits constant intensity.
  • AI and analytics pipelines: Throughput bursts over modest datasets. Model Flexible against Ultra.
  • VDI and file shares: Moderate, steady demand. Premium or Standard first; Flexible if capacity dominates.
  • Backup or DR: Capacity-heavy, idle performance. Flexible or Standard.
  • Dev/test: Temporary and uneven. Flexible with short-term clones.

Decision Checklist

  1. Measure current utilization.
  2. Identify peak throughput.
  3. Confirm capacity growth.
  4. Map protocols and availability zones.
  5. Check regional quotas.
  6. Model pricing.
  7. Test representative traffic.
  8. Set a review cadence.

Eight-step Azure NetApp Files service level decision checklist

Performance, Limits, and Capabilities to Validate

Service level isn't the only factor. Volume size, large-volume eligibility, client VM network limits, protocol and mount configuration, concurrency, and workload mix all shape what you observe.

Limits to Know

Limit Value
Capacity pool 1-2,048 TiB (4 TiB minimum with Basic network features)
Regular volume 50 GiB-100 TiB
Large volume 50 TiB-1 PiB by default
Large-volume throughput ceiling 12,800 MiB/s
Pools per account 100
Volumes per pool / subscription 1,000 / 2,000

Most limits can be raised by support request. Microsoft's table specifies no fixed per-tier IOPS cap.

Capabilities That Drive Decisions

  • Protocols: NFSv3, NFSv4.1, SMB, and dual-protocol. Dual-protocol requires an Active Directory connection.
  • Replication: Cross-region (asynchronous) and cross-zone for DR designs.
  • Snapshots and backup: Cover recovery objectives and dev/test cloning.
  • Encryption: AES-256 at rest, with platform- or customer-managed keys.
  • Identity: Microsoft Entra Kerberos is preview and SMB-only, so don't plan NFS around it.

Validating Performance

Microsoft's benchmark on a 101-TiB Ultra volume with 12 clients reached roughly 12,761 MiB/s sequential reads and 709,000 random-read IOPS. That reflects a specific setup, not a guarantee.

Microsoft's testing guidance warns that one client, one network flow, or one file won't show full capability. Test sequential throughput, random I/O, metadata-heavy operations, and latency separately. For Linux NFS, Microsoft suggests nconnect=4 in most environments, and advises against combining it with sec=krb5*.

How to Evaluate and Manage ANF Service Levels

Right-Sizing Workflow

  1. Inventory pools and volumes.
  2. Collect utilization and throughput telemetry from Azure Monitor. Track Total throughput, Total IOPS, latency, and Throughput limit reached (equals 1 when QoS throttling occurs).
  3. Separate peak from sustained demand.
  4. Map volumes to applications.
  5. Compare provisioning with measured needs.

Azure NetApp Files right-sizing workflow from telemetry to application mapping

Planning a Change Safely

Use Change Pool to move a volume in place to a pool in the same NetApp account. Microsoft documents it as requiring no data migration and not affecting access. Plan around these rules:

  • Upward moves have no wait; a downgrade requires waiting at least 24 hours after an upward move.
  • Flexible pools can't be converted to or from other pool types.
  • Moving to a manual-QoS pool keeps existing throughput until you edit it.
  • Resource paths change, so old logs and metrics don't follow. Recreate diagnostic settings.
  • Test with a non-production workload and prepare rollback steps.

Governance

  • Assign ownership across ITOps, DevOps, and FinOps.
  • Review cost and performance together.
  • Tag by application and environment.
  • Alert on sustained underutilization, saturation, and unexpected throughput growth.

Teams running Azure managed disks or other block storage alongside ANF may find Lucidity useful. It offers multi-cloud block-storage visibility and autonomous optimization across AWS, Azure, and Google Cloud. Lucidity isn't an ANF service-level management tool and doesn't replace Azure NetApp Files.

Frequently Asked Questions

What are Azure NetApp Files service levels?

Standard, Premium, Ultra, and Flexible are capacity-pool options that set how performance and capacity are provisioned. Check Microsoft documentation for current limits and availability.

What is the difference between Standard, Premium, Ultra, and Flexible?

The first three use fixed profiles of 16, 64, and 128 MiB/s per TiB. Flexible lets you select capacity and throughput independently.

How does the Flexible service level work?

You provision pool capacity and throughput separately. Each pool includes 128 MiB/s, extra throughput is added per MiB/s, and decreases follow a 24-hour cooldown.

Which service level is best for high-performance workloads?

Ultra fits consistently high demand tied to capacity. Flexible fits high throughput on small datasets or variable peaks. Validate with telemetry and current limits.

Can you change the service level after deployment?

Yes. Use Change Pool to move a volume to another pool in the same NetApp account. Downgrades require a 24-hour wait after an upward move, and Flexible pools cannot be converted to another service level.

Does the Flexible service level reduce costs?

It can, when capacity and throughput needs diverge. Microsoft's examples show 25% to 39% savings, but model your own usage against current US Azure pricing.