IT and cybersecurity · solution

Fault-tolerant cluster on HyperSphere (ГиперСфера) software

A replacement for VMware vSphere and Stratus everRun.Russian software built on KVM virtualisation for continuously available fault-tolerant clusters; it replaces Stratus everRun and VMware vSphere. It installs on two servers, and if one fails the protected virtual machines migrate live to the standby server automatically. Management is through a web interface.

Servers2VM failoverautomatic, FT or HAVirtualisationKVMSpecialistscertified
Server 1VM · control systemVM · archiveServer 2 · standbyVM · control systemVM · archiveautomatic migrationmanagement through a web interface
Two servers: if the first fails, the protected virtual machines migrate live to the standby server

Three properties

  1. 1

    Reliability

    continuous operation of the protected virtual machines: if one server fails they migrate live to the standby server

  2. 2

    Competitiveness

    a counterpart to Stratus everRun and VMware vSphere

  3. 3

    Security

    compliance with Russian information protection law

How a two-server cluster works

A cluster built on HyperSphere (ГиперСфера) high-availability clustering software by STR LLC consists of two physical servers that operate as a single system. According to the developer, a protected virtual machine runs on both servers at the same time: if one server fails, the machine keeps running on the other, and the failure is transparent to the applications inside it.

The solution is based on checkpointing. The state of the guest virtual machine is replicated to the second server as checkpoints; if a failure occurs, the standby server continues from the current checkpoint. Data and memory are synchronised over direct Ethernet links between the servers, so no separate storage system, SAN or Fibre Channel is required.

The protection mode is set for each virtual machine individually. The scheme described above is fault tolerance (FT). In high availability (HA) mode, the machine is restarted on the second server if its server fails. One cluster can host both machines that must never stop and less critical services.

Virtual machine protection modes in HyperSphere
CriterionFault tolerance (FT)High availability (HA)
If a server failsthe machine keeps running on the second server without a reboot or stopthe machine is restarted automatically on the healthy server
Interruption for applicationspractically nonefor the time it takes to restart the machine and its applications
Typical useservers whose downtime interrupts process control or accounting: SCADA, databasesservices that can tolerate a restart

Characteristics

HyperSphere software characteristics
CharacteristicHyperSphere
Virtualisation typeKVM
Number of servers in the cluster2
VM migration on failureautomatic: in FT mode without stopping the machine, in HA mode by restarting it on the standby server
Managementweb interface
Compliance with Russian information protection lawyes

Platform and product names are used to identify the systems we service. Trademarks belong to their respective owners. Energy Management LLC is an independent engineering company.

Server and network requirements

The cluster requires two x86-64 servers with Intel or AMD processors. HyperSphere is built on KVM, the hypervisor that is part of the Linux kernel and relies on Intel VT or AMD-V hardware virtualisation. The developer recommends fully identical servers with compatible processors, both running the same operating system.

According to the developer, a cluster hosts up to 24 virtual machines, and a machine in FT mode can have up to 8 vCPUs. Protected machines can run GNU/Linux, Microsoft Windows and FreeBSD; Russian operating systems such as RED OS and Astra Linux are supported.

Cluster networks

  • direct Ethernet links between the servers (Private and A-link), used to synchronise data and memory contents; in the developer’s reference configurations they run at 10 Gbit/s or more;
  • connections to the corporate network, which carry user traffic to the virtual machines and cluster management through the web interface.

The servers are sized to the number and load of the virtual machines and to their protection mode. Processor resources and memory are reserved not only for the machines themselves but also for the HyperSphere services that make them fault-tolerant.

Reference cluster server configurations according to the developer
LoadVirtual machinesServer
Low1–3 in HA mode or 1 in FT modeone entry-level processor (4–12 vCPUs), 32 GB RAM, 1 TB on SATA/SAS drives with hardware RAID
Medium4–6 in HA mode or 1–3 in FT modetwo mid-range processors (16–24 vCPUs), 64 GB RAM, 2–4 TB on SATA/SAS drives with hardware RAID
Highmore than 6 in HA mode and/or up to 8 in FT modetwo high-end processors (36 vCPUs or more), 128 GB RAM, 4–8 TB on SSDs (SAS or NVMe) with hardware RAID

Which systems go into the cluster

Servers whose downtime stops a process or the business are moved into the fault-tolerant cluster. Applications inside the protected machines need no changes: fault tolerance is provided by the cluster, not by the application.

  • process level: SCADA and HMI servers, process data historians, telecontrol and equipment monitoring systems, i.e. the server level of a process control system;
  • production and accounting: MES, ERP, the 1C:Enterprise server and its DBMS, other database servers;
  • general IT infrastructure: domain controller, terminal, file and print servers;
  • site security: access control systems, video surveillance and building management systems.

For workshops, power facilities, mines, quarries and other sites without a server room, the developer describes a cluster on two passively cooled industrial servers that can be installed directly in an electrical cabinet.

Energy Management uses HyperSphere to build fault-tolerant clusters as part of integrated solutions for the power sector, industry and other sectors.

Migrating from VMware vSphere and Stratus everRun

According to the developer, virtual machines are moved to HyperSphere without data loss: from VMware vSphere with standard import and export tools, and from Stratus everRun live, without stopping them. Migration from Microsoft Hyper-V is also supported.

How the migration runs

  1. Inventory: a list of the virtual machines, their resources, guest operating systems and links to other systems.
  2. Choice of protection mode for each machine and sizing of the two servers.
  3. Compatibility check of guest operating systems and applications; no more than 8 vCPUs for machines in FT mode.
  4. Migration of the machines and a failover test: one server is shut down while the applications are running.
  5. Staff drills for failure scenarios and an update of the operating documentation.

A move to Russian virtualisation platforms starts with an IT infrastructure audit, one of the company’s IT and cybersecurity services.

Fault tolerance in the requirements for significant CII objects

For significant critical information infrastructure (CII) objects of categories 2 and 1, redundancy is part of the basic set of security measures. The set is defined by the Requirements for Ensuring the Security of Significant CII Objects approved by FSTEC of Russia Order No. 239 of 25 December 2017 (measure codes transliterated from the Order):

  • ODT.1: use of fault-tolerant hardware;
  • ODT.2: redundancy of equipment and systems;
  • ODT.3: monitoring of the failure-free operation of equipment and systems;
  • DNS.4: redundancy of software, hardware and communication channels in case of emergencies.

Measure ODT.7, clustering of the information (automated) system, is not in the basic set for any category: it is added when the basic set is not enough to neutralise the threats to the object.

Since 1 January 2025, government bodies and customers purchasing under Federal Law No. 223-FZ (except organisations with municipal participation) have been prohibited from using foreign software on their significant CII objects, unless otherwise established by federal law (Presidential Decree No. 166 of 30 March 2022). The virtualisation platform for such objects is therefore chosen from Russian software; HyperSphere is developed by the Russian company STR LLC.

Security measures for significant objects are designed together with the server architecture; see cybersecurity for industrial control systems. The company holds FSTEC of Russia licence No. L024-00107-77/01590240 of 4 December 2024 for the technical protection of confidential information; see licences and approvals.

Certified specialists

In November 2025 the company's specialists were certified in HyperSphere administration (news of ).

Partner

HyperSphere (ГиперСфера) high-availability clustering software by STR LLC; the developer publishes no official Latin spelling of the name. Energy Management LLC is a HyperSphere software partner under an agreement of ; all partnerships.

Questions and answers

What happens if one server fails?

Machines in fault tolerance (FT) mode keep running on the second server without a reboot, so the applications are not interrupted. Machines in high availability (HA) mode are restarted automatically on the healthy server.

How does HyperSphere differ from VMware vSphere and Stratus everRun?

It is Russian software built on KVM virtualisation and compliant with Russian information protection law; in purpose it is a counterpart to everRun and vSphere for continuously available clusters.

How many servers are needed?

Two: the cluster installs on a pair of servers and is managed through a web interface.

Who builds HyperSphere clusters?

Energy Management LLC is a partner of the developer STR LLC (agreement of 2025); our specialists are certified in HyperSphere administration (November 2025).

Does a HyperSphere cluster need a storage system?

No. According to the developer, the cluster needs no storage system, SAN or Fibre Channel: two x86-64 servers with their own drives, connected by direct Ethernet links, are enough. The data and memory of the protected virtual machines are synchronised over these links, so if one server fails, the other holds an up-to-date copy of them.

Can a 1C server be placed in the cluster?

Yes. The 1C:Enterprise server and its DBMS run in protected virtual machines like any other server application, with no modifications. The developer lists 1C servers and database servers among the typical HyperSphere use cases. For a database whose downtime interrupts accounting, fault tolerance (FT) mode is chosen if 8 vCPUs are enough for the machine: if a server fails, it keeps running without a reboot.

Can the cluster servers be placed in different buildings?

Yes. According to the developer, HyperSphere nodes can be deployed behind separate firewalls, on different floors, in different buildings or in different parts of a city, up to 10 km apart. Such a geographically distributed cluster uses synchronous replication: if one site fails, the services are restarted automatically at the other. The links between the sites must be able to carry the data synchronisation.

Engineering team

IT and cybersecurity

Platform and product names are used to identify the systems we service. Trademarks belong to their respective owners. Energy Management LLC is an independent engineering company.

CallSend a request
Service requestRequest a quotationData sheet