High availability – following the backup rule

by Billy J Long Guidance, PSA Articles

Article (PSA‑0005)

What “High Availability” (HA) Really Means

High Availability is a design goal that ensures a system delivers an agreed level of uptime –‑ usually 99.9 % (≈ 8 h downtime/yr) or higher –‑ even when components fail. Modern businesses (hospitals, data‑centers, SaaS providers, remote offices) depend on HA to keep critical applications running 24/7.

Key HA concepts (2025)

  • Redundancy – Duplicate hardware or virtual instances (servers, storage, network paths) so a single failure never stops service.
  • Failover & Automatic Switchover – Monitoring detects a failure and instantly routes traffic to a standby component (e.g., active‑passive cluster, hot‑standby VM).
  • Load Balancing – Distributes traffic across multiple nodes, improving performance and providing another layer of fault tolerance.
  • Geographic Distribution – Deploying services across multiple data‑center locations or cloud regions reduces the impact of site‑wide outages.
  • Replication & Data Synchronisation – Keeps data copies in near‑real‑time (block‑level or file‑level) on separate nodes.
  • RPO & RTO – Recovery Point Objective (how much data loss is tolerable) and Recovery Time Objective (how quickly service must be restored). HA architectures are built to meet the RPO/RTO goals you define.

Where Backups Fit In

Backups are the foundation of any HA strategy, but they’re just one piece of the puzzle. A solid backup plan protects you from data loss caused by hardware failure, ransomware, accidental deletion, or catastrophic events.

The classic 3‑2‑1‑0 rule (still the gold standard)

  1. 3 – Three copies – Primary data + two additional backups.
  2. 2 – Two media types – For example, an internal NAS (or disk) plus cloud object storage (e.g., Backblaze B2, Azure Blob, Amazon S3).
  3. 1 – One off‑site location – Store at least one copy in a different physical site or a cloud region.
  4. 0 – Zero‑error verification – Test restores regularly (at least quarterly) to confirm backups are usable.

2025‑enhanced backup practices

  • Immutable storage – Write‑once, read‑many (WORM) or object‑storage lock features that prevent even administrators from overwriting recent backups. This thwarts ransomware that tries to encrypt backups.
  • Snapshot‑based protection – Use volume snapshots (VSS on Windows, LVM snapshots on Linux, or ZFS) for near‑instant point‑in‑time copies.
  • Hybrid cloud backup – Combine on‑premise fast restores with cloud durability; many solutions now offer built‑in encryption, compression, and bandwidth throttling.
  • Automated backup testing – Scripts that periodically restore a random file or database row and verify checksum integrity.
  • Ransomware‑aware backup policies – Separate “live” backup streams from “archival” immutable copies; rotate the live backups daily, weekly, monthly.

Putting It All Together: A Simple HA Blueprint

  1. Assess critical services. Identify which applications, databases, and file shares must stay online.
  2. Define RPO/RTO targets. Example: RPO = 15 minutes for ERP database; RTO = 30 minutes for web portal.
  3. Build redundancy. Deploy two servers (or VMs) in an active‑active cluster behind a load balancer; add a second network path (dual ISP or VLAN).
  4. Implement replication. Use real‑time mirroring (e.g., Storage‑Space‑Direct, DRBD, or cloud‑native database replication) to keep data in sync across nodes.
  5. Apply the 3‑2‑1‑0 backup rule. Schedule daily incremental backups + weekly full backups, store one copy locally, one copy in a second media type, and one copy in a secure cloud region.
  6. Test failover and restore. Quarterly, simulate a server loss and verify that traffic switches automatically, then run a backup‑restore drill to validate data integrity.
  7. Monitor and alert. Use an RMM or SIEM to watch health metrics, backup job success, and latency; set up alerts for any breach of RPO/RTO.

Common Pitfalls to Avoid

  • Treating backups as a “set‑and‑forget” task – without regular testing, backups can be corrupted or incomplete.
  • Relying on a single backup media type (e.g., only external hard drives) – hardware failures are inevitable.
  • Storing all copies in the same physical location – a fire or flood can wipe everything.
  • Neglecting encryption and access controls – unprotected backups are a gold mine for attackers.
  • Ignoring the human factor – document procedures, train staff, and enforce least‑privilege access to backup systems.

Bottom Line

High Availability is much more than “just a backup”. It blends redundancy, real‑time replication, automated failover, and rigorous testing to keep services running. The 3‑2‑1‑0 backup rule remains the foundation, but in 2025 you should augment it with immutable storage, cloud snapshots, and regular restore verification to meet modern RPO/RTO expectations.

Need a Custom HA & Backup Strategy?

PSA Computer Services can design, implement, and test a solution that meets your uptime goals and budget.

Call us today at (707) 506‑6802 for a free assessment.