Selectel and Hystax launch managed disaster recovery into the cloud

On 15 June 2026, Selectel and Hystax announced a disaster-recovery service that uses Hystax Acura to recover infrastructure into Selectel’s cloud. It is designed to replicate virtual machines, applications, data and network configuration with a prepared failover procedure.

What the primary source confirms

The service maintains an up-to-date infrastructure copy in a standby Selectel region. The source can be the customer’s own virtual environment or another provider. The vendors also state that recovery can work across different hypervisors, a relevant feature for hybrid and multi-platform estates.

The managed service includes:

• recovery architecture design;
• data-replication configuration;
• prepared recovery workflows;
• test failovers that simulate primary-site unavailability;
• separate parameters for systems of different criticality;
• standby capacity in Selectel’s cloud platform.

It complements Selectel’s Multi-AZ region but addresses another layer. Availability zones mitigate failures inside a cloud region, while this DR workflow moves a workload from another infrastructure estate into the recovery environment.

DR is more than a backup

A backup preserves data but does not define the order in which databases, queues, APIs and external integrations restart. A DR plan connects copies to compute capacity, networking, DNS, secrets, dependencies and the eventual return to the primary site.

The service’s value therefore depends on achievable recovery objectives:

• RPO defines the acceptable amount of recent data loss;
• RTO defines the acceptable service-recovery time;
• a test failover demonstrates that the runbook works in practice.

Set objectives per business service. Applying the shortest possible RPO and RTO to every system can add substantial cost and complexity without a matching benefit.

Questions for a pilot

1. Which hypervisors, operating systems and disk types are supported?
2. How are crash-consistent and application-consistent datasets handled?
3. What bandwidth is required for the target RPO?
4. Who changes DNS, routes, VPNs and external allowlists?
5. How are keys, certificates and secrets recovered without unsafe copying?
6. How are dependencies started and data consistency verified?
7. Can an isolated test run without affecting production?
8. How does failback work after the primary site is restored?
9. Which metrics and reports demonstrate RPO and RTO?
10. How is responsibility divided between the customer, Selectel and Hystax?

For SMS, email, payment gateways and other external APIs, decide how the recovery environment obtains network access and current integration settings. A restored VM is not useful if a partner does not allow its address or its credentials are obsolete.

Remaining limitations

The official announcement does not publish universal RPO/RTO values, pricing, a complete version-compatibility matrix or independent test results. These depend on data volume, change rate, network capacity and application architecture.

The cross-hypervisor claim must be tested for the exact platform pair. Drivers, disk formats, boot configuration, network interfaces and licensing may require extra work. A representative pilot should reproduce the service topology rather than migrate one isolated test VM.

Practical conclusion

The joint service turns DR from a collection of custom scripts into a managed workflow with recovery tests. It may suit organisations with on-premises virtualisation or several providers that need a common recovery environment in a Russian cloud.

Start by mapping dependencies and classifying services, then recover one medium-criticality environment and measure the actual result. See our disaster recovery plan and backup and fault-tolerance guide for the wider methodology.

Source: Selectel’s official announcement dated 15 June 2026.