Understanding VMware Site Recovery Manager: Architecture, Hybrid Cloud Disaster Recovery, and Comparison with Azure Site Recovery

In the previous article we discussed Microsoft Azure Site Recovery (ASR) for cloud‑to‑cloud disaster recovery; this piece focuses on VMware's Site Recovery Manager (SRM) solution and briefly touches on the vSAN Stretched Cluster feature.

VMware SRM is very similar to Microsoft ASR: both offer hypervisor‑level and storage‑level replication, require storage‑vendor providers (SRA) for integration, and support disaster‑recovery scenarios between private clouds and between private and public clouds.

VMware provides HA, FT, vMotion, Storage vMotion, Data Recovery, and VADP backup interfaces for VM continuity, and now adds SRM and vSphere Replication, which correspond to Microsoft’s ASR and Replica respectively.

The architecture (see image below) shows that SRM supports both array‑based replication and native vSphere Replication. Array remote replication leverages the array’s own replication features together with an SRA plugin, while vSphere Replication tracks changed disk blocks and copies only incremental data.

Hybrid Cloud Disaster Recovery uses vSphere Replication for data copy, as explained in the previous article. Because vSAN does not provide built‑in remote replication, vSphere Replication is typically used for vSAN‑based storage. Unlike Microsoft ASR, which requires creating an Azure account and launching recovery tasks via the Azure portal, SRM can be deployed at two sites, synchronizing VM configuration information automatically for high reliability.

SRM’s architecture based on vSphere Replication follows a client‑server model: the protected site runs an agent to send data, while the recovery site runs a Replication Server that receives the data.

Private Cloud Disaster Recovery relies on storage‑array replication to synchronize data between sites. SRM uses a storage vendor’s SRA (similar to Microsoft’s ASR Provider) and requires the vendor to supply a vCenter plugin or VASA driver for discovery, configuration, and management of storage resources. This approach supports both synchronous and asynchronous replication, enabling zero‑data‑loss protection across two data centers.

Comparing the two methods, vSphere Replication simplifies DR management by being a native VMware component that works per‑VM but consumes host resources, whereas storage‑based replication offloads the copy work to the array (via SRA) but only supports LUN‑level protection.

At the end of the article we introduce the vSAN Stretched Cluster, VMware’s dual‑active solution built on a cross‑site distributed vSAN cluster combined with HA, RDS, and other high‑availability features. It is part of the vSphere Metro Storage Cluster (vMSC) certification, which defines requirements for data‑transfer distance, networking, and arbitration. In vSphere 6.0, VMware added the SRM Stretch feature, which uses SPBM policy management to replace the older vMSC certification.

Scenario Suitability

SRM works with relatively relaxed requirements for distance, network bandwidth, and site configuration, offering managed, consistent workflows, testing capabilities, manual steps, custom processes, and reporting to ensure correct disaster‑recovery execution.

vSAN Stretched Cluster provides active‑active, zero‑downtime capabilities between two sites located within roughly 100 km of each other, with high‑bandwidth, low‑latency links, delivering near‑zero RTO and RPO and automatic failover.

Note: Scan the QR code below or search for “ICT_Architect” to follow the public account for more content.