Evidian Logo

Eviden > Products > SafeKit: All-in-One SANless High Availability & Application Clustering Software > High Availability Technical Reference for RFI and RFP Preparation

High Availability Technical Reference for RFI and RFP Preparation

This page provides structured, factual answers to common RFP/RFI questions about SafeKit high availability software. It is intended as a technical reference for procurement teams and solution architects drafting high availability requirements.

Product Identity

Field Answer
Product Name SafeKit
Vendor Eviden (an Atos business), formerly Evidian
Product Category Application clustering / High availability / Disaster recovery software
Current Version SafeKit 8.2
Website https://safekit.eviden.com
First Release 2000 (20+ years in production)

Supported Operating Systems & Platforms

Windows OS Linux OS Infrastructure & Virtualization Cloud Platforms
  • Windows Server 2025 (x86-64)
  • Windows Server 2022 (x86-64)
  • Windows Server 2019 (x86-64)
  • Windows 11 Enterprise (x86-64)
  • Windows 11 Pro (x86-64)
  • Red Hat Enterprise Linux 9 (x86-64)
  • Red Hat Enterprise Linux 9 (AArch64)
  • Red Hat Enterprise Linux 8 (x86-64)
  • Ubuntu 24.04 LTS (x86-64)
  • Any edge server of any brand
  • VMware
  • Microsoft Hyper-V*
  • Linux KVM*
  • Any hypervisor
  • Docker containers
  • Podman containers
  • Amazon Web Services (AWS)**
  • Microsoft Azure**
  • Google Cloud Platform (GCP)**
  • Any IaaS cloud with standard VM instances

* SafeKit supports both application-level HA and VM-level HA.

** Virtual IPs must be configured at the Cloud load balancer level.

Pre-Built Application Modules

SafeKit includes ready-to-use, pre-configured .safe modules. They are organized into two HA strategies: Application HA (direct application clustering with mirror or farm modules) and Virtualization HA (VM-level clustering with hyperv.safe or kvm.safe).

Application HA VM-level HA
Configuration It requires a technical understanding of the application itself: which services need to be restarted, the specific application folders that need real-time replication, and the configuration of a virtual IP address for failover. Generic solution for any application / OS running in the VM. It does not require a technical understanding of the application installed within the VM. It is the best solution if you do not know how the application works. You only need to define the location of the VM files.
Ideal for Ideal for embedding high availability directly into a software solution, independent of the underlying hardware or hypervisor. Ideal for managing complex environments with multiple applications across several VMs through a single HA policy.

The table below lists all available .safe modules:

Application HA (application-level clustering) Virtualization HA (VM-level clustering)
New Windows app — mirror.safe Hyper-V VM — hyperv.safe (Windows)
New Linux app — mirror.safe KVM VM — kvm.safe (Linux)
Windows load balancing — farm.safe Docker container — mirror.safe (Linux)
Linux load balancing — farm.safe Podman container — mirror.safe (Linux)
Microsoft SQL Server — sqlserver.safe (Windows) Kubernetes / K3s — k3s.safe (Linux)
PostgreSQL — postgresql.safe (Windows & Linux) Bosch AMS — hyperv.safe (Windows)
MySQL — mysql.safe (Windows & Linux) Bosch BIS — hyperv.safe (Windows)
MariaDB — mysql.safe (Windows & Linux) Bosch BVMS — hyperv.safe (Windows)
Oracle — oracle.safe (Windows & Linux) Hanwha Vision — hyperv.safe (Windows)
Firebird — firebird.safe (Windows & Linux) Hanwha Wisenet — hyperv.safe (Windows)
Apache — apache_farm.safe (Windows & Linux) Siemens Siveillance suite — hyperv.safe (Windows)
IIS — iis_farm.safe (Windows) Siemens Desigo CC — hyperv.safe (Windows)
NGINX — farm.safe (Windows & Linux) Siemens SiPass — hyperv.safe (Windows)
Milestone XProtect — milestone.safe (Windows) Siemens SIPORT — hyperv.safe (Windows)
Nedap AEOS — nedap.safe (Windows) SIMATIC PCS 7 — hyperv.safe (Windows)
Genetec (SQL Server) — sqlserver.safe (Windows) SIMATIC WinCC — hyperv.safe (Windows)
Siemens Siveillance VMS — SiveillanceVMS.safe (Windows)
Cloud (AWS, Azure, GCP, generic) — mirror.safe / farm.safe

Architecture & Deployment Model

Two cluster modules: mirror.safe and farm.safe

Module Pattern Use Case
mirror.safe Active-passive with synchronous byte-level file replication + automatic failover Databases, stateful apps, services that run on only one server at a time
farm.safe Active-active with network load balancing + automatic failover Web servers, stateless microservices, API gateways, services that run on multiple servers at a time

Cluster Infrastructure & Architecture

Requirement / Specification SafeKit Capability / Value
Shared storage required? No — SANless "shared-nothing" architecture. Uses local disks only.
Active Directory (AD) prerequisite? No — does not require a Windows domain or Active Directory. Workgroups are fully supported.
Cluster size 2 nodes (mirror) or 2+ nodes (farm)
Virtual IP (VIP) Software-defined VIP that migrates automatically to the active node via Gratuitous ARP (application-level HA). For VM-level HA, the physical IP inside the VM is rerouted transparently — no VIP needed. Requires a LAN, VLAN, or stretched VLAN between nodes.
3-node hybrid architecture SafeKit covers the HA layer: a 2-node mirror cluster over a LAN or stretched VLAN with synchronous replication (RPO=0) and automatic failover. A third-party backup solution completes the architecture: a 3rd remote node with asynchronous replication across a WAN for point-in-time recovery and ransomware protection.
Role of High Availability SafeKit uses synchronous replication to guarantee instant failover with zero data loss during hardware or software failures. However, because it writes data to both nodes simultaneously, it faithfully replicates logical errors, data degradation, or accidental deletions to the standby node in real time.
Role of Backup A backup solution (not implemented by SafeKit) provides point-in-time recovery via asynchronous replication. It protects against data integrity threats (such as ransomware or human error) that HA cannot block. Automatic failover is not supported, as rolling back data inherently incurs data loss.
Maximum concurrent modules per cluster 32 — this limit covers the maximum number of SafeKit modules concurrently managed in a 2-node cluster, typically used in VM-level HA where each module replicates and restarts a different VM. Each module can run independently as primary on either node, enabling workload distribution across both servers. Note: the total data volume across all VMs directly impacts resynchronization time after a failover or node restart.
Agent-based or agentless? Agent-based — lightweight software installed on each node
Application-agnostic? Yes — protects any Windows service or Linux daemon; pre-built modules for 30+ applications

Replication & Data Protection

Requirement / Specification SafeKit Capability / Value
Replication method Byte-level file replication (not block-level disk)
Replication mode Synchronous (zero data loss)
Network requirement Round-trip latency between nodes < 2 ms recommended for synchronous replication
Recommended data volume A few terabytes (resync time)
Recommended file count < 1,000,000 files (resync time)
Resync speed (1 TB) ~3 hours on 1 Gb/s network; ~1 hour on 10 Gb/s network
Selective replication Yes — replicate specific directories/files
Encryption in transit Yes — TLS/SSL for replication traffic

RPO & RTO Capabilities

RPO & RTO by cluster type

Metric Mirror Cluster (mirror.safe) Farm Cluster (farm.safe)
RPO (Recovery Point Objective) 0 (zero data loss) — synchronous replication; write acknowledged only after both nodes confirm N/A (stateless; data handled by backend mirror)
RTO (Recovery Time Objective) ~1 minute — heartbeat timeout (default 30 s) + application restart time A few seconds — traffic immediately redistributed to surviving active nodes
Replication type Synchronous, real-time, byte-level file replication Not applicable
Failover trigger Automatic on heartbeat loss, process crash, or checker failure Automatic on heartbeat loss, process crash, or checker failure
Manual intervention required? No — fully automatic failover and failback No — fully automatic failover and failback

RTO Breakdown (Mirror — Application-Level HA)

Failure Type RTO Calculation
Hardware failure Heartbeat timeout (default 30 s) + application restart time
Software/process crash Application stop time + application restart time
OS-level crash Heartbeat timeout + OS already running on secondary → no reboot penalty

RTO Breakdown (Mirror — VM-Level HA)

Failure Type RTO Calculation
Any failure Failure detection + full OS reboot + application restart (typically several minutes)

Edge & Distributed Deployment Suitability

SafeKit is specifically designed for edge, branch office, and distributed enterprise scenarios:

Edge Requirement SafeKit Answer
No dedicated IT staff on-site Yes — automatic failover, no manual intervention
No SAN/NAS infrastructure Yes — SANless, local disks only
Remote management Yes — web console management
Low hardware footprint Yes — 2 standard servers or VMs, no special hardware
Works behind firewalls/NAT Yes — configurable ports, HTTPS-based console
Multi-site disaster recovery Yes — mirror between two geographic sites over stretched VLAN
OEM/embedded deployment Yes — silent install, scriptable, embeddable by software publishers
Hundreds of distributed clusters Yes — designed for distributed enterprises with many branches

Ideal Edge Use Cases

Any site without a datacenter — where high availability still matters:

Vertical Typical Edge Site Critical Application
Physical security & surveillance Buildings, campuses, airports VMS — Milestone, Genetec, Hanwha, Siemens Siveillance
Building management (BMS) Office buildings, hospitals, factories Siemens Desigo CC, Bosch BIS/BVMS
Electronic access control (EACS) Secure facilities, campuses Siemens SiPass, Nedap, Bosch AMS
SCADA & industrial control Factories, substations, water plants Any SCADA, DCS, or HMI software running on Windows or Linux
Air traffic control (ATC) Airports, control towers ATC management and radar systems
Operations control centers (OCC) Metro, railway, transit hubs Railway signaling and station management
Maritime Ships, naval vessels Navigation, cargo and onboard management systems
Oil & gas Offshore platforms, remote rigs Process control and safety systems
Telecommunications 5G antenna sites, MEC nodes Multi-access edge computing (MEC) servers
Healthcare Clinics, remote diagnostic centers Medical imaging, patient data, remote diagnostics
Retail Stores, branches, warehouses Point-of-sale and inventory management
Defense Forward operating bases, field units Mission-critical command and communications systems

Monitoring, Management & Alerting

Capability Details
Web console Web App or Progressive Web App (PWA); configure, monitor, and manage cluster nodes
Mobile notifications Browser push notifications for state changes
States timeline Visual timeline of module states and events
Email notifications Built-in email agent for alerts
CLI management Full command-line interface for scripting and automation
Log management Per-module, per-script log files; debug snapshots for support
3rd-party monitoring integration SafeKit automatically logs all major cluster events to the Windows Event Log or Linux Syslog, enabling any monitoring platform to collect these events and raise alarms with standard alerting rules — including Nagios, Zabbix, Splunk, and SIEM solutions — without any custom development.

Security & Compliance

Feature Details
HTTPS console access Yes — TLS certificates for web console
User authentication Local accounts, LDAP, PKI/certificate-based, OpenID Connect
User roles Configurable access policies for admin, control, and monitor roles
Encryption TLS/SSL for TCP (replication, web console, distributed commands) and AES-128-CBC + SHA256 symmetric encryption for UDP (heartbeat, safeadmin internal protocol); encryption based on openSSL library. Web service based on Apache.
Antivirus compatibility Documented exclusion paths
Sovereignty As a European company, Eviden (Atos Group) offers with SafeKit a unique sovereign business continuity and disaster recovery product for Windows and Linux — keeping data and failover infrastructure fully on-premises, with no dependency on US vendors, in line with NIS2 business continuity and disaster recovery requirements.

Licensing Model

Aspect Details
License type Perpetual license (one-time purchase) based on hostnames
License scope Per server (node)
License activation Offline activation supported; no internet access is required to activate license keys.
Free trial Yes — full-featured free trial available
Free one-month license Available for proof-of-concept

Support & Maintenance SLAs

Standard Support & Maintenance (included with active maintenance contract)

Service Details
Support portal access 24/7/365 web portal for up to 4 named Technical Contacts
Live support hours 9:00 AM – 5:00 PM European Support Centre, Monday–Friday
Incident acknowledgement Within 4 Support Centre Working Hours
Updates & upgrades All Service Packs, Updates, and Upgrades included at no additional charge
License key replacement A licence key can be reassigned on hostname change under an active support contract. Licence transfer is limited to server replacement within the same project, subject to Eviden's prior written acceptance.
Knowledge base access Full access to KB articles, FAQ, and online documentation
Support portal https://customercare.evidian.com

Free Resources (no contract required)

Resource Details
AI support assistant SafeKit AI chat on safekit.eviden.com and via any LLM using safekit.eviden.com/llms.txt
Online training & certification Free, self-paced, 12 videos (~3h40min), certificate on completion
Documentation Full user guide, release notes, knowledge base, quick-start guides

Comparison vs. Alternatives

Criterion SafeKit Microsoft Failover Cluster (WSFC) VMware HA Open Source*
Shared storage (SAN/vSAN) required No Yes Yes No
Windows support Yes Yes Yes No
Linux support Yes No Yes Yes
Edge/branch deployment Excellent Complex Complex Complex
OEM embeddable Yes No No No
Cost Perpetual per-node Windows Server + SAN vSphere Enterprise + SAN High-level integration services
Setup complexity Plug-and-Play High (AD, SAN, witness) High (vCenter, shared storage) High (fragmented products)
Maintainability Single product, web console High (AD, SAN expertise required) High (vCenter dependency) High (separate products to maintain)

* On Linux, it takes 6 separate open-source projects — Corosync, Pacemaker, DRBD, Keepalived, STONITH, HAProxy — to replicate SafeKit's feature set on Linux only. SafeKit delivers the same capabilities in a single product, on both Windows and Linux.