Architecture patterns

These are the deployment patterns Datamotive supports in production. Every pattern uses the same building blocks — Management Server pairs, Replication Node pairs, and recovery-site Prep/DeDup Nodes — arranged per the deployment models.

Pattern 1 — On-premises VMware to cloud DR

A vSphere environment replicates to AWS, Azure, or GCP, and protected VMs recover as native cloud instances.

• Management Server on both the vCenter site and the cloud region; Replication Node pairs added per 40 protected disks.
• Agentless, block-level replication driven by vSphere CBT; Windows recoveries use a Prep Node in the cloud region.
• Inter-node traffic runs over private connectivity or VPN; nodes need outbound 443 to vCenter and the cloud APIs. See Networking.

Pattern 2 — Cloud-to-cloud DR

Workloads in one cloud replicate to another cloud (or another region), removing single-provider dependence. Supported sources are VMware, AWS, and Azure; targets add GCP — see the platform matrix.

• Datamotive nodes are deployed in both clouds, in the same regions as the source and recovery workloads.
• Enable compression and optionally a DeDupe Node to cut cross-cloud egress; the dashboard reports the achieved data reduction.
• Plan cloud quotas before scale events — see Limits.

Pattern 3 — Multi-AZ recovery

Recoveries spread across availability zones in the target region for resilience:

• One Management Node in any zone (Azure: deploy with the No Zone option to enable multi-zone recovery).
• A Prep/Replication Node in each AZ where recoveries are expected — each supports 20 parallel VM recoveries in its zone. See Preparing target site.

Pattern 4 — Migration cutover

The same replication foundation performs one-way migrations: replicate while the source runs, then cut over with zero data loss — manually (power off, final zero-change iteration, migrate) or via Auto-Migrate, which orchestrates the sequence. A DeDupe Node is particularly effective for one-time migrations with similar workload datasets. See Run a migration.

Pattern 5 — Isolated drill environment

Alongside any DR pattern, a sandboxed infrastructure on the recovery site — VPCs, networks, and subnets with no outbound connectivity to production or DR — lets test drills run with zero risk. Drills recover from the last consistent replicated copy without touching production or replication, and don't consume license capacity. See DR drill.

Related docs

• Deployment models

• Preparing target site

• Scale and performance guide