There is no single Azure Storage IOPS limit. The applicable target depends on the service and the resource being measured: an Azure Files share or file, a storage account, a Blob object, or a VM disk. Microsoft publishes different targets for each, and actual performance depends on workload and access patterns. Identify the service, tier, region, and resource scope before comparing your workload with any number.
Azure Storage IOPS targets at a glance
The figures below are Microsoft-published service targets, not independent benchmarks or promises that an application will achieve that rate. “Requests per second,” “transactions per second,” and IOPS can describe different operations and scopes; they are not interchangeable workload guarantees. The figures reflect the Microsoft guidance summarized as of September 28, 2026, and published targets can change.
As an Amazon Associate I earn from qualifying purchases.
| Service and scope | Published target | How to interpret it |
|---|---|---|
| Azure Files, provisioned v2 SSD share | 3,000 minimum and 102,400 maximum provisioned IOPS per share | Share-level provisioned IOPS. The same guidance gives a separate maximum of 12,000 data IOPS for an individual SSD file; share and file figures describe different scopes. |
| Standard Azure Storage account | 40,000 requests per second in the regions listed by Microsoft’s target page; 20,000 requests per second in regions not listed | An account request-rate target, not a guaranteed application IOPS rate. Check the live Microsoft page for which regions are in its named set. |
| Azure Blob Storage, individual object | Up to 3,000 requests per second for one block blob; up to 500 requests per second for one page blob | Per-blob request targets. They do not describe the whole account, and a hot partition can affect performance before an account-wide target is reached. |
| VM unmanaged disks using a standard storage account | 20,000 IOPS maximum total request rate per standard storage account | Microsoft says total IOPS for unmanaged VM disks in that account should not exceed this rate. Do not apply this account-specific guidance to all managed disk SKUs. |
Which IOPS number applies to your workload?
Azure Files: share provisioning and per-file performance are separate
For a provisioned v2 SSD share, Microsoft lists a minimum of 3,000 and a maximum of 102,400 provisioned IOPS at the share level. It separately lists up to 12,000 data IOPS for an individual SSD file. A share’s provisioned target does not mean each file can reach that same rate: account and share limits still matter, as do I/O size and queue depth. Microsoft recommends SSD shares for Azure Files workloads that need high IOPS, fast transfer, or low latency.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Storage account request rate: region matters
Microsoft’s standard storage account target is 40,000 requests per second in its named set of regions and 20,000 requests per second in regions outside that set. This is a request-rate target for an account, not a statement that any workload will achieve the same number of application IOPS. Because the regional grouping can change, check the current target page for the applicable region rather than assuming a rate from another deployment.
#1 Best Overall
Blob Storage: account, blob, and partition are different scopes
For an individual blob, Microsoft lists targets of up to 3,000 requests per second for a block blob and up to 500 requests per second for a page blob. Those per-object figures do not replace the storage account’s own scale targets. Nor do they rule out localized contention: a hot partition can experience latency and return HTTP 500 or 503 errors before the account as a whole approaches its target.
VM disks: keep the unmanaged-disk qualification
Microsoft’s VM disk guidance describes a maximum total request rate of 20,000 IOPS for a standard storage account and says the combined IOPS of unmanaged VM disks in that account should stay within it. This is specifically an unmanaged-disk/account consideration; it should not be generalized into a universal limit for managed disk SKUs.
Why a published IOPS target may not match observed performance
IOPS means input/output operations per second, but the count alone does not describe the work performed. A workload issuing many small requests differs from one issuing fewer large requests. Read/write mix, operation type, concurrency or queue depth, access pattern, throughput, client capacity, and network distance can all affect the result. Microsoft’s Azure Files guidance calls out I/O size and queue depth; its storage resource-provider guidance says achieved request rate and bandwidth depend on object size, access patterns, and workload.
Microsoft uses the word “target” rather than treating every published value as an absolute limit because some values may be increased upon request. Approaching or exceeding a target can still lead to throttling and higher latency. Treat the published figure as a planning reference for its stated service and scope, then validate with representative traffic.
Rank #3
How to diagnose an Azure Storage IOPS problem
- Pin down the resource and configuration. Record the service, SKU or tier, region, redundancy option, and whether the target applies to an account, share, file, blob, partition, or disk. Do not compare an individual-object target with an account-wide measurement.
- Measure representative application traffic. Capture request or I/O size, read/write mix, concurrency or queue depth, latency, throughput, and request errors. Include ordinary operation and the bursts or peak periods that reproduce the problem.
- Look for localized throttling. Check whether latency or HTTP 500 Operation Timeout and HTTP 503 Server Busy responses point to a hot partition or a broader account/service ceiling. Errors can appear at a partition before account-wide targets are reached.
- Change one relevant factor at a time. If the evidence indicates partition contention, distribute traffic across partitions where the design permits. If bursts are causing pressure, smooth them instead of sending sudden spikes.
- Retest after each change. Compare the same workload characteristics and observe both throughput and latency, not just the peak request count. A design option or a higher published target does not by itself establish end-to-end application performance.
Ways to improve performance, matched to the bottleneck
- For Azure Files workloads: consider an SSD share when high IOPS, fast transfer, or low latency is required, and check both account/share limits and per-file behavior.
- For Blob workloads with hot partitions: distribute traffic across partitions when possible, avoid abrupt spikes, and use exponential backoff after HTTP 503 responses so retries do not intensify the load.
- For Blob workloads requiring higher transaction rates or lower latency: evaluate a premium block blob account. Hosting the storage account in the same region as clients can reduce network latency, but neither option guarantees a particular end-to-end IOPS rate.
- For custom Blob applications: use Microsoft Storage client libraries; Microsoft says they are optimized and incorporate established performance practices.
What to compare before choosing a tier or redesigning storage
Compare like with like. A useful performance assessment records the service and resource scope, tier and provisioning model, region and redundancy configuration, and the relevant account-, share-, file-, blob-, partition-, or disk-level targets. Include throughput and per-object limits where applicable, then describe the workload in terms of request size, read/write mix, concurrency, latency needs, and client location. If a target depends on regional availability or may be increased on request, verify that condition against Microsoft’s current documentation.
For published figures, use Microsoft’s service-specific target documentation: Scalability and performance targets for Azure Files, Scalability targets for standard storage accounts, Scalability and performance targets for Blob Storage, Scalability and performance targets for the Azure Storage resource provider, Performance checklist for Azure Blob Storage, and Microsoft’s VM disk performance guidance. These are Microsoft guidance figures, not measured typical results.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Free tools Windows power users keep installed
One-click scans. No signup required.
