High Availability and Disaster Recovery (HADR) Software without Shared Disks

How to implement HADR over a fast network?

To implement HADR over a fast network, deploy two nodes across remote sites to provide High Availability and Disaster Recovery simultaneously. This setup ensures that your application remains available even if an entire site fails, without the complexity of a shared storage infrastructure.

Why use a SANless HADR solution for fast networks?

A SANless solution like Evidian SafeKit is essential because it replicates data synchronously and in real-time between two nodes without shared disks. Because synchronous replication ensures zero data loss (RPO=0), the software can trigger an automatic application failover instantly if a hardware or software failure occurs.

What are the network requirements for Synchronous HADR?

Synchronous HADR requires an extended LAN or a stretched VLAN for three critical reasons:

• Virtual IP Failover: A single subnet is required to automatically failover the virtual IP address between nodes.
• Low Latency: Synchronous replication with no data loss typically requires a network round-trip time of less than 2ms.
• High Bandwidth: A connection of 1 Gb/s or more is required to ensure rapid data resynchronization during a failback.

How to implement HADR over a slow network?

To implement HADR over a slow network (WAN), the architecture typically separates High Availability from Disaster Recovery. In this scenario, two nodes are deployed at the primary site for local redundancy, while a separate backup or asynchronous replication strategy is used to protect data at a distant remote site.

What is the best HADR solution for low-bandwidth connections?

How does failover work with a remote Disaster Recovery site?

Because asynchronous replication over a slow network involves potential data loss (RPO > 0), the failover process to a disaster recovery site is typically manual and managed by an administrator:

• Data Restoration: Backups are restored onto secondary servers at the DR site. Modern solutions like Veeam are often used to restore virtual machines (VMs) quickly to reduce recovery time.
• DNS Rerouting: Traffic is redirected to the DR site at the DNS level. The recovery time depends on DNS cache timeouts (TTL), and some client applications may require a restart to pick up the new IP address.
• Manual Decision: An administrator must verify data integrity at the remote site before authorizing the switch to the secondary data center.

Comparing High Availability and Data Backup

Although both are critical for data protection, High Availability (HA) and Backup Solutions target different risks. High Availability, powered by SafeKit, provides a "live" failover mechanism to keep applications accessible during server outages or hardware failures. Conversely, a Backup Solution acts as a "historical" archive. While HA ensures 99.99% system availability by replicating data in real-time, backup focuses on data integrity, providing the necessary restoration points to recover from logical errors, accidental deletions, or ransomware attacks.

Is High Availability a substitute for a Backup strategy?

No, High Availability and backups are complementary, not interchangeable. While SafeKit ensures business continuity by keeping applications running during a hardware crash, it does not guard against logical errors, accidental deletions, or ransomware attacks. For example, because real-time replication mirrors every change instantly, a ransomware attack on the primary node will be immediately duplicated on the secondary node. To recover from such cyber threats or accidental deletions, you need a dedicated backup solution with a robust retention policy. This allows you to "rewind" your environment to a healthy state from before the corruption occurred.

Optimizing RTO and RPO: The Synergy Between HA and Backup

• High Availability (via SafeKit): Targets Instant Recovery. By maintaining a live, synchronous mirror of your environment, SafeKit achieves near-zero RTO and zero RPO. If a server fails, the application resumes on the secondary node immediately, ensuring no service interruption or data loss during the transition.
• Backup Solutions: Target Data Resilience. While HA keeps the service "live," backup creates an "immutable" history. It is your fallback for when the live data itself is compromised, allowing you to restore a clean version of your database from a specific point in time before a virus or accidental deletion occurred.