Introduction
Data is the single most valuable asset your business owns. Lose it, and you could lose months of work, customer trust, and regulatory compliance. Yet most growing businesses treat storage and backup as an afterthought β "we'll figure it out when we need it."
By then, it's usually too late.
At Hostingowy, we manage petabytes of storage across our infrastructure. This guide covers the storage and backup architecture we've built for ourselves and our enterprise customers β from NVMe performance tiers to ZFS snapshot strategies and cross-region replication.
The Three Pillars of Enterprise Storage
A robust storage strategy rests on three pillars: performance, durability, and recoverability. Each requires different technology choices and trade-offs.
Pillar 1: Performance β Choosing the Right Storage Tier
Not all data needs to be on the fastest storage. A tiered approach saves money while delivering performance where it matters:
| Tier | Hardware | Use Case | Typical Latency |
|---|---|---|---|
| Tier 0 β Hot | NVMe (PCIe Gen4/5) | Databases, active websites, application servers | <0.5ms |
| Tier 1 β Warm | SATA SSD / NVMe | File storage, staging environments, logs | 1-5ms |
| Tier 2 β Cold | HDD / Object Storage | Archives, backups, compliance data | 10-50ms |
At Hostingowy, our Storage plans start at $30/month for NVMe-backed hot storage with automatic snapshot capabilities. For cold storage and archival, we provide S3-compatible object storage at $0.005/GB/month.
Pillar 2: Durability β ZFS and Data Integrity
We use ZFS as our primary filesystem across all storage infrastructure. ZFS offers several critical advantages for data durability:
- Copy-on-write β Data is never overwritten in place, preventing corruption from write failures
- Checksumming β Every block is checksummed, detecting (and auto-repairing) silent data corruption that bit rot causes
- RAID-Z β Software RAID that outperforms hardware RAID controllers and supports online expansion
- Snapshots β Instant, space-efficient point-in-time snapshots with near-zero performance overhead
- Compression β Built-in LZ4 compression (typically 1.5-3x savings) with negligible CPU cost
For enterprise customers, we configure ZFS with RAID-Z2 (double parity) β tolerating up to two concurrent drive failures without data loss.
Pillar 3: Recoverability β The 3-2-1 Backup Rule
The 3-2-1 backup rule is the gold standard: 3 copies of your data, on 2 different media types, with 1 copy off-site. Here's how we implement it:
Our Backup Architecture
Layer 1: ZFS Snapshots (On-Server)
Automated ZFS snapshots provide the first line of defence. We take:
- Hourly snapshots β Retained for 24 hours
- Daily snapshots β Retained for 14 days
- Weekly snapshots β Retained for 4 weeks
- Monthly snapshots β Retained for 3 months
Snapshots are space-efficient β they only consume storage for data that changes between snapshots. A typical daily snapshot uses only 0.5-2% of the original volume size. Restoring from a snapshot is instant: just zfs rollback tank/data@snapshot-name.
For managed backups through our platform, see our backup documentation.
Layer 2: Off-System Backups (Separate Storage)
Snapshots protect against accidental deletion or corruption, but not against a full server failure. For that, we replicate snapshots to separate storage servers using zfs send/receive:
# Automated incremental replication
zfs send -R -I tank/data@daily-0 tank/data@daily-1 \
| ssh backup-server "zfs receive -F backup-pool/data/"This sends only the changed blocks between snapshots β typically 50-200 MB per daily increment for a 100 GB volume. Replication runs over encrypted SSH tunnels to dedicated backup storage nodes.
Layer 3: Off-Site / Cross-Region Backups
For maximum durability, critical data is replicated off-site:
- Nightly rsync to a second European data centre (geographic redundancy within Europe)
- Weekly encrypted backups to S3-compatible object storage
- For compliance requirements: immutable backup storage with WORM (Write Once, Read Many) policies
Cross-region replication adds an RPO (Recovery Point Objective) of 24 hours and an RTO (Recovery Time Objective) of 4 hours for full restore.
Backup Schedule Recommendations
| Data Type | Snapshot Frequency | Retention | Off-Site Frequency |
|---|---|---|---|
| Production databases | Hourly | 72 hours hourly, 30 days daily | Continuous WAL streaming |
| Web application files | Daily | 14 days | Daily |
| Customer data | Daily | 90 days | Daily |
| Logs & analytics | Weekly | 90 days | Weekly |
| Compliance archives | Monthly | 7 years | Monthly |
Testing Your Backups (The Most Overlooked Step)
A backup that hasn't been tested is a wish, not a backup. We recommend:
- Weekly automated restore tests β Restore a random snapshot to a staging server and verify data integrity
- Monthly full restore drills β Simulate a total server failure and measure your team's ability to restore from scratch
- Quarterly disaster recovery exercises β Full DR simulation including communications, escalation, and post-mortem
Our platform includes automated backup validation β every backup is tested immediately after creation with cryptographic integrity verification.
Conclusion
A proper storage and backup strategy is not optional β it's foundational to running a serious business online. Start with the 3-2-1 rule, implement ZFS snapshots for instant recovery points, and always test your backups.
At Hostingowy, every plan β from $5/month VPS to enterprise dedicated servers β includes ZFS-based storage with automated snapshots. Sign up today and get your first month free with code LAUNCH100. Or contact our sales team to discuss a custom storage解ε³ζΉζ‘ for your business.