IBM Storage Virtualize in PBHA+DR deployments

Hello IBM Team,

We would like to submit a Request for Enhancement for IBM Storage Virtualize in PBHA+DR deployments.

In our environment, we have implemented a 3-site replication topology based on  PBHA+DR architecture, where two storage systems form the HA pair and one is in the DR location, as follows:

System 1 - production site and actively serves host I/O.
System 2 - located in a clean-room environment (in synchronous replication distance) and, by design, has no host attachment.
System 3 - disaster recovery location (asynchronous replication distance).
In this topology, the current split-brain handling mechanism can lead to undesirable outcome in certain failure scenarios.

Specifically, if the Active Management System (AMS) role is held by the clean-room system at the moment when communication between the two HA systems is lost, the production system is forced to take the partition volumes offline.
As a result, the production site loses access to data, and the running applications fail, despite the fact that it is the only site actively serving host I/O.
From both operational and business continuity perspective, this behavior is undesired for deployments where one HA member is intentionally hostless and should not take precedence over the production system for I/O continuity decisions.

The current AMS-based model provides a solid and well-structured foundation for high-availability scenarios, particularly in environments where fully automated and uniform failover decisions are appropriate.

However, in our specific deployment model, we believe there would be of significant value to complement this behavior with an optional, customer-selectable policy that allows split-brain authority to be aligned with the operational priorities of the environment.
Such an enhancement would extend the flexibility of the solution for specialized topologies, while fully preserving the existing AMS-based behavior for deployments where it is already the optimal choice.

Examples of possible implementation approaches could include:

Allowing customers to explicitly designate a Preferred Management System to be selected as the split-brain winner.
Allowing customers to define a policy that keeps the system currently serving active host I/O during split-brain conditions.

Implementing this capability would provide the following benefits for our deployment:

Improved continuity of production application services.
More predictable behavior during communication loss scenarios, where a human decision must be taken depending on the sites communications availability.
Better alignment with air-gapped clean-room architectures.
Reduced risk of unnecessary production outages in such a configuration.
We respectfully request that this proposal be considered as a potential product enhancement for future releases of IBM Storage Virtualize.

Kind regards,
Virgil Balan
Storage Administrator
CEC Bank SA