Skip to content

Azure Elastic SAN Next Generation public preview: Mission-critical performance & cloud-native scale

Mission-critical applications such as SAP HANA, Epic, and core transactional databases demand consistently high performance and predictable low latency. Cloud-native platforms running stateful applications, vector databases, and fast-growing fleets of AI agent environments need persistent storage that can scale across thousands of volumes. Meeting these growing storage demands without overprovisioning resources or adding operational complexity is a common challenge that organizations face today.


To address this challenge, we are announcing the public preview of Azure Elastic SAN Next Generation, bringing together the performance mission-critical applications demand, the scale cloud-native workloads need, and the cost efficiency enabled by Elastic SAN’s storage consolidation model and dynamic performance sharing. Elastic SAN Next Generation expands the workloads that can benefit from Elastic SAN—whether you’re migrating an on-premises SAN estate, modernizing enterprise applications, or building cloud-native applications.


At a glance




  • Mission-critical performance. Up to 400,000 IOPS and 10,000 MB/s per volume, with low sub-millisecond latency and noisy-neighbor protection for Business Critical volume groups.

  • Cloud-native scale. Up to 1 PiB of capacity, 4 million IOPS, 160 GB/s of throughput, and 20,000 volumes per SAN.

  • Cost efficiency. Thin provisioning, autoscale, thin cloning (coming soon), and independent scaling of capacity and performance, built on the consolidation economics of Elastic SAN.

  • Data protection and recovery. Instant snapshots with automatic hardening, complemented by application-consistent backups through Commvault and Druva.



Elastic SAN Next Generation builds on the Elastic SAN foundation customers already know: a familiar SAN, volume-group, and volume hierarchy; enterprise-grade security, data protection and resiliency; and shared storage infrastructure that lowers storage TCO and simplifies storage management.


Mission-critical performance for your most demanding applications


For I/O-intensive enterprise applications, storage performance directly affects how quickly transactions complete and how predictably applications respond under load. Consolidation is valuable only when the storage platform can meet those requirements.


The new Business Critical tier delivers up to 5× the IOPS and nearly 8× the throughput per volume compared with today’s General Purpose iSCSI volumes, alongside low sub-millisecond latency. Quality of Service (QoS) provides noisy-neighbor protection at the Business Critical volume-group level, protecting configured IOPS and throughput allocations from contention elsewhere on the SAN.


For customers migrating an on-premises SAN estate, moving data is only part of the challenge—the destination must also meet the different service levels their applications depend on. Business Critical and General Purpose bring that familiar tiered-storage approach to Azure, with higher performance and noisy-neighbor protection extending Elastic SAN to more demanding applications. This helps customers map workloads across their SAN estate to appropriate storage tiers and plan storage migrations at the SAN level—helping accelerate SAN migrations to Azure during datacenter exits and hardware refreshes.


These applications also benefit from the economics of a shared SAN. Customers can reserve performance for critical workloads while sharing available performance across volumes and provisioning capacity to match their needs—helping improve utilization and reduce overprovisioning.


For the representative workload configuration evaluated, Elastic SAN Next Generation offers approximately 43% lower monthly storage costs than competing cloud provider. See “Consolidation economics in practice” below for more details.


Scale persistent storage for cloud-native and AI workloads


Growing SaaS database fleets, distributed vector databases, and persistent agent stores and workspaces can create demand for thousands of volumes. Across these workloads, customers need storage that keeps pace while remaining simple to provision, organize, and manage.


Elastic SAN Next Generation supports up to 20,000 volumes per SAN at preview, with higher limits on the roadmap. For database fleets and agent platforms using persistent block storage on Azure, this provides room to expand through a familiar SAN management model. Volumes remain organized into groups, with network and security policies applied centrally rather than configured volume by volume—bringing scale and management simplicity together as application estates grow.


Shared performance helps improve utilization as activity varies across databases and agent environments, while thin provisioning and autoscale help align capacity provisioning with growth. Thin cloning (coming soon) will provide space-efficient, instantly mountable volume copies for agent workspaces that start from a common baseline—helping teams expand these environments without making a full data copy for each one.


More workloads. Greater control. Better storage economics.


Across these workloads, Elastic SAN Next Generation helps customers match performance and connectivity to their applications, provision resources efficiently as demand grows, and protect data with appropriate recovery options.


Choose the performance and connectivity that fit your workloads


Assign a QoS tier to each volume group: Business Critical offers reserved IOPS and throughput for noisy-neighbor protection and can draw on available shared performance beyond its reservation. General Purpose uses dynamically shared performance without reservations.


The new Direct Attach option combines a familiar disk-attachment workflow for supported Azure virtual machines, with SAN-level provisioning and volume group management. Customers don’t need to configure guest-side iSCSI sessions.


iSCSI remains available to provide low-cost external storage for bare-metal environments, including Azure VMware Solution (AVS) and Nutanix, and for any workload that needs high volume density per host. Take agent sandboxes as an example. If each agent gets its own isolated volume, a single VM can need 100+ volumes. iSCSI handles that scale and still delivers the raw performance of block storage.


QoS tier

Connectivity

Performance model

Max IOPS per volume

Max throughput per volume

Example workload profiles

Business Critical

Direct Attach

Volume-group performance reservations with noisy-neighbor protection; low sub-millisecond latency

400,000

10,000 MB/s

Transaction-heavy databases and demanding enterprise applications

General Purpose

Direct Attach

Shared performance; sub-millisecond latency

80,000

2,000 MB/s

Vector databases, open-source databases, persistent application stores

General Purpose

iSCSI

Shared performance

80,000

1,280 MB/s

Azure VMware Solution, Nutanix, and other iSCSI-preferred environments


Per-volume performance depends on volume capacity, any reservations on the Volume Group and the performance provisioned at the SAN level.


Lower storage TCO as your workloads grow


Performance sharing improves utilization. Flexible provisioning complements it by helping customers keep capacity and performance spending aligned with what their applications need:



  • Thin provisioning: give applications room to grow without buying all that capacity upfront. The aggregate logical capacity assigned to volumes can be up to 5× the SAN’s provisioned capacity. Capacity charges remain based on the SAN capacity provisioned—not the full logical capacity assigned to volumes.

  • Autoscale: reduce manual capacity management. Automatically grow SAN capacity based on utilization thresholds you define, helping accommodate growth without repeatedly resizing capacity yourself.

  • Independent provisioning: buy capacity and performance according to your needs. Provision capacity, IOPS, and throughput independently at the SAN level, rather than purchasing additional capacity solely to obtain more performance.

  • Thin cloning (coming soon): create new copies without duplicating all the data upfront.Create space-efficient, instantly mountable copies of Business Critical volumes for development, testing, and agent workspaces that start from a common baseline.


Consolidation economics in practice


Consider a workload portfolio of mission-critical databases, cloud-native workloads and their application servers, requiring 150 TiB of logical volume capacity to accommodate 60 TiB of current data and projected near-term growth. Thin provisioning allows customers to present the full logical capacity to their VMs, while initially provisioning and paying for 66 TiB of SAN capacity: 60 TiB for current data and 6 TiB of headroom with autoscaling enabled. The model also assumes workload peaks do not fully coincide: their combined peak requires 400,000 IOPS & 12.8 GBps, compared with 600,000 IOPS & 19.2 GBps when provisioning each workload independently for its peak.


Resources and cost

Elastic SAN Next Generation

Other Cloud Provider

Total logical volume capacity

150 TiB

150 TiB

Current data usage

60 TiB

60 TiB

Provisioned capacity

66 TiB

150 TiB

Provisioned IOPS

400,000

600,000

Provisioned throughput (MBps)

12,800

19,200

Modeled monthly storage cost

$25,764.05

$45,536.30


At this initial utilization, the modeled monthly storage cost is approximately 43% lower with Elastic SAN Next Generation. The savings reflect capacity provisioned for current data (including headroom), included baseline performance of 2000 IOPS & 64 MBps per provisioned TiB, and dynamic performance sharing across workloads whose peaks do not fully coincide. As data usage grows, autoscale increases provisioned SAN capacity up to the configured maximum of 150 TiB, with capacity charges increasing accordingly.


Enhanced Data Protection and Recovery


As more applications rely on shared storage, recovery becomes an essential part of the storage strategy. Elastic SAN Next Generation combines native snapshots with partner-enabled backup and recovery, supporting both rapid volume recovery and application-aware protection.


Instant snapshots with automatic hardening. Incremental, crash-consistent snapshots provide instant access and instant restore for five hours after creation. In parallel, each snapshot is automatically hardened to lower-cost storage. Once hardened, its lifecycle is independent of the source volume, preserving a recovery point even if that volume is deleted. Hardened snapshots support standard restores, with an option for faster restores.


Application-consistent backups through leading partners. Leading third-party backup partners such as Commvault and Druva build on Elastic SAN snapshots to provide enterprise-grade protection, including application-consistent backups, advanced backup policies and retention management, and recovery workflows ranging from cross-region disaster recovery to granular file and folder restore. Integration with Veeam is on the roadmap – Capabilities and supported scenarios vary by integration.


Get started with the public preview


Azure Elastic SAN Next Generation is now in public preview with locally redundant storage (LRS) in France Central and Southeast Asia. UK South, East US and West US 2 are coming soon.


Start with a representative workload and evaluate the tier, connectivity, and provisioning model that fit your environment. Review the documentation for supported VM configurations, connectivity options, and preview limitations before you begin.



Whether you’re modernizing the applications that run your business or building the platforms that come next, Azure Elastic SAN Next Generation expands your storage choices—with performance matched to your workloads, room to grow, and a familiar model for keeping storage efficient and manageable.


Mission-critical performance. Cloud-native scale. The consolidation economics of Elastic SAN.


We can’t wait to see what you build.

Microsoft Tech Community originally posted this article on 18 November 2026 at 9:14 AM.

Leave a Reply