Understanding Component Redundancy
Key Takeaways
Cloud service continuity measures and customer connector recovery protect different components.
VA cluster redundancy depends on connected members that can reach both the cloud and the relevant target.
Direct SaaS connectors have no customer VA failover path.
Element health and a recovered business operation require different evidence.
Trace the operations in the deployment
A browser or application reaches the tenant service through its documented interface. The platform processes requests and can communicate with external applications through the configured connector. For a VA-based source, the Cloud Connector Gateway (CCG), running on eligible members of the configured virtual appliance (VA) cluster, handles the connection path. The target directory or application, its credentials and its network remain distinct dependencies. A SaaS connector works from a cloud-managed connector path and does not rely on a customer VA cluster. Identify which components an operation uses before choosing what should fail over.
The SailPoint Security Addendum describes provider business continuity and disaster recovery controls, including independent cloud availability zones, backups, restoration testing, and recovery planning. A recovery time objective describes how long service restoration is intended to take in the applicable plan. A recovery point objective describes the intended bound on data loss expressed as a point in time. Backups and redundancy solve different failure modes. An alternate runtime can sustain some operations, while restoring persisted state addresses data loss or corruption. The public description of controls does not supply elapsed recovery results for a specific tenant incident. For a named deployment, use the applicable service documentation and terms rather than assuming cross-region replication or a recovery guarantee for every dependency.
| Component | High-availability or recovery focus | Evidence during restoration |
|---|---|---|
| Managed cloud service | Provider continuity and disaster recovery plan | Region and service status, then success of a relevant tenant operation |
| Customer VA cluster and CCG | Multiple paired members with enough capacity and complete network paths | Healthy members, source association and source operation results |
| Standby VA location | Configured cross-site or replacement design | Reachable target, actual source path and queued workload results |
| SaaS connector target | Connector service, API credentials and target's own availability | Connector activity and native application status |
| External directory, database or trusted service | Target service and its network, authentication and recovery plans | Native health plus a controlled operation through the intended connector |
Work failures with an unaffected comparison
A planned rotation restarts one VA node. A second member is connected to the same cluster and can reach the tenant's cloud service and payroll directory. Other work for that cluster can continue through the healthy member. SailPoint's rolling update documentation still notes that a long-running task can be interrupted. The engineer checks the actual aggregation record, start and end data, and results before declaring the workload healthy. SailPoint requires at least two members per cluster for this deployment and lists no requirement for a customer load balancer. In a controlled exercise, record actual throughput with one member unavailable and check whether each active member can reach the dependent targets. Verify available processing headroom when a member is unavailable. Two VAs do not remove a shared dependency if both rely on a single route or unavailable target server.
Now assume both VAs are healthy and a different connected directory works, but payroll fails following an expired service credential. Additional VAs or a move to the backup cluster would leave that credential failure in place. Test the payroll connector and repair the affected credential or target path with the source owner. If the primary site is down and the organization preconfigured a separate VA cluster at a recovery site, follow its switch-clusters procedure. That involves reassigning affected sources and reentering credentials. A cross-site design using active reachable members of one cluster can continue through its surviving members without that source rebinding, assuming the target is still reachable.
Verify after the switch
Begin with a small, supported source operation, then confirm the work that users care about. For a read path, check an approved aggregation sample and its timestamp. For a write, capture account-operation status and confirm the specific native effect when necessary. A green status on a cloud dashboard or a connected VA member can verify a component. A recovered connector test, job, and target outcome show whether that evidence extends to the business requirement. Record the failure boundary and dependencies that remain common to the replacement path.
See SailPoint’s Security Addendum section 8.6 and 8.7, VA resiliency designs, VA management, and source types.
A VA rolling update restarts one node in a two-member cluster. Normal connectivity remains through its peer. What proves that a particular aggregation survived?
The peer displays Connected.
The aggregation record shows the expected completed result and the required data is present.
The tenant uses cloud availability zones.
The maintenance window elapsed.
A direct SaaS connector returns an expired target credential error. Its native API is reachable. Which recovery action targets the failing dependency?
Move the source to a backup VA cluster.
Change the tenant portal URL.
Increase the number of active VA members.
Correct the credential used by that SaaS connector and verify the target operation.
Sections you finish are checked off in the contents.