Concept
Replication Patterns
The policy layer for controller-backed replication and disaster-recovery target readiness.
Definition
A replication pattern is the Sendense policy that defines disaster-recovery replication behavior: which workloads are replicated, where their recovery-side replicas live, how recovery readiness is maintained, and how failover workflows are coordinated.
- Which workloads participate in the replication workflow.
- The recovery site and destination platform, with placement settings such as template, network, disk offering, service offering, and zone where applicable.
- The replication network for target-side replication traffic, the recovery network used when a workload boots during failover, and optional test network overrides.
- The sync interval, presented as the pattern RPO target, and whether manual full resync is available.
- The replication route policy and pattern-level sync concurrency and queuing behavior.
Patterns And Targets
A replication pattern is the policy. A replication target is the per-workload target created under that policy. Each target tracks the source workload, destination placement, controller VM, sync schedule, last sync state, and failover lifecycle state.
Operators can trigger a pattern-wide sync that fans out across the assigned targets. Sendense starts some syncs immediately, queues others when concurrency limits are reached, skips targets that are not syncable, and reports where a sync request cannot be accepted.
Controller-Backed Targets
When a replication pattern uses controller-backed targets, Sendense provisions one controller VM per protected VM after the required controller template is available in the destination system.
Replication Route Policy
Each pattern carries a replication route policy that controls how replication data travels between the protected site and the recovery site. Automatic lets Sendense choose the best route for each sync, Via SHA sends data through the encrypted appliance connections via SHA, and Direct P2P sends it directly between the two site SNAs after a live reachability check.
Sync Lifecycle
The pattern's sync interval is the target RPO for its workloads. RPO status is an operational measurement based on last successful sync and current target state, not a guarantee under every condition.
The initial full sync establishes the destination replica and its per-disk baselines. Later full or incremental syncs update that same controller-backed target in place.
Before a destination update can modify a controller-backed replica, Sendense prepares a short-lived rollback checkpoint. A successful sync removes it; an interrupted sync that may have written data is recovered to the prior disk state and replication baseline before normal syncs resume.
Sync lifecycle
Known state
Last Successful Replica
The destination replica disks and Sendense replication baselines describe the same completed sync.
Before writes
Prepare Checkpoint
Sendense prepares the destination rollback state and records the matching pre-sync replication baselines before the update begins.
Target update
Apply Sync Changes
The target SNA and Sendense Controller apply the next full or incremental update to the replica disks.
Success path
Commit And Clean Up
Sendense commits the new baselines, deletes the temporary checkpoint, and updates readiness and RPO status.
Target update
Apply Sync Changes
The target SNA and Sendense Controller apply the next full or incremental update to the replica disks.
Failure path
Revert And Recover
If writes may have started, Sendense reverts the destination disks and restores their previous replication baselines before another sync is allowed.
Known state
Last Successful Replica
The destination replica disks and Sendense replication baselines describe the same completed sync.
Pattern-Level Failover
Replication patterns support failover for a single VM or across the pattern. A pattern failover processes the eligible targets, can run multiple VMs in parallel, and lets operators limit parallelism and choose whether to continue with remaining VMs if one VM fails.
Test failover keeps rollback available. Planned failover powers down the source where possible, runs a final sync, and starts the recovery-side workloads. Emergency failover starts recovery-side workloads from the latest available sync point when the source is unavailable. After a planned or emergency failover, the operator commits to finalize or rolls back while rollback remains available.
Replication Patterns And Protection Patterns
Protection patterns and replication patterns are both policies orchestrated by SHA, but they serve different goals. A replication pattern is not a backup retention policy and does not store backup data.
Related Docs