Disaster Recovery & Business Continuity
12
Disaster Recovery & Business Continuity
Backups protect data. Availability protects access. Disaster recovery is what happens when both get tested for real — this field guide covers backup strategy, replication, resource availability, and the RTO/RPO math that drives site selection.
$ dr-status --check rto rpo replication
# cross-referencing backup jobs, AZ health, failover targets
✓ 3-2-1 rule satisfied · ✓ RPO 1h met · ✓ hot site synced
$
Backup Types
Every backup strategy is a trade between how long the backup job takes and how long the restore takes. The four core types make that trade differently.
Beyond the big three
- Synthetic full — assembled from the last full plus subsequent incrementals, not pulled from live data
- Snapshot — read-only, point-in-time; common for VMs, Active Directory, Exchange (e.g. AWS EBS Snapshots)
- Archive — not a backup at all; optimized for cheap long-term retention, not protection (AWS Glacier, Azure Archive Storage, GCP Nearline/Coldline/Archive)
Backup Objects & Locations
What you back up — and where you put it — depends on how often it changes and how fast you'd need it back.
| Object | Protects | Typical frequency |
|---|---|---|
| System-state | Core OS configuration | On major config/version change |
| Application-level | App configs, executables, customizations | Before version migration |
| Filesystem | User data | Every 24 hours (typical) |
| Database dump | DB contents via SQL statements | Multiple times a day for busy DBs |
| Configuration file | Ansible/Chef/Puppet/DSC files | Any time edited |
The 3-2-1 rule
3 copies of data, on 2 different media types, with 1 copy offsite. With cloud storage, a common modern variant keeps a single onsite copy and replicates two cloud copies across different regions.
Backup Targets & Schedules
Tape, disk, and cloud each solve a different part of the cost-versus-speed-versus-control equation.
| Target | Strength | Trade-off |
|---|---|---|
| Tape | Cost-to-capacity, long-term archive | Slow to access |
| Disk | Fast backup and restore | Not ideal for long-term archive |
| Cloud / object storage | Low upfront cost, scalable, globally accessible | Internet connection becomes a speed variable |
Schedule guidelines
- User data — frequent enough to recover file versions (often daily)
- Configuration files — any time a change is made
- System-state — on major version or config changes
- Database data — multiple times a day for busy databases; database settings themselves rarely change
Data Retention & Replication
Retention governs how long data must — and can — be kept. Replication keeps copies current across servers and sites.
Retention best practices
- Maintain compliance with industry regulation and legal requirement minimums and maximums
- Track data sovereignty considerations for where retained data lives
- Classify data first — classification drives retention rules
- Automate deletion and purging to stay compliant and control storage cost
Encryption & Recovery Testing
Encrypted backups protect confidentiality. Regular restore testing is what proves recoverability actually works.
| Encryption scope | Protects against |
|---|---|
| In transit | Interception while moving across LAN, WAN, or to the CSP |
| At rest | Exposure of data sitting on storage media, private or public cloud |
Testing levels
- Individual file restore — simulates accidental user deletion
- System recovery — simulates a server failure
- Disaster recovery — simulates a widespread outage
Recovery Types & Options
The right restore type depends on whether you're recovering for real or just testing that recovery works.
| Restore type | When to use it |
|---|---|
| In-place / overwrite | Real incident — roll back to a known point, overwriting current data |
| Side-by-side / parallel | Compare backed-up files against current production without overwriting |
| Alternate location | Testing — restore to a separate system (often a VM) while production stays live |
| Granular file | Recover individual modified or deleted files |
| Snapshot restoral | Recover a VM to a specific point in time; also used to migrate data across AWS accounts/regions |
Resource Availability
Backups protect data that's already gone. Availability architecture tries to keep things from going down at all.
Regions & availability zones
Each cloud region contains multiple availability zones — data centers with isolated power and networking, connected by fast links, but far enough apart to avoid sharing a single point of failure.
Availability techniques
- Cloud bursting — overflow to public cloud only when on-prem/private capacity is exceeded
- Edge computing — move compute near the data source (e.g. IoT sensors) instead of shipping data to a central site
- NIC teaming — combines NICs into one virtual NIC for failover and link aggregation
- Load balancing — distributes connections across servers; round-robin, most-recently-used, or weighted scheduling
| Monitoring tool | Type |
|---|---|
| Prometheus / Grafana | Open-source |
| AWS CloudWatch / Azure Monitor / GCP Monitoring | CSP-native |
Disaster Recovery & Sites
RTO and RPO are the two numbers that actually decide which site tier you can afford — drag the sliders to see how they trade off.
DR site tiers
| Site type | Cost | Data handling | Equipment present |
|---|---|---|---|
| Hot | High | Immediate replication | All |
| Warm | Medium | Data migrated | Some |
| Cold | Lower | Data migrated | None |
| Cloud | Variable | Variable | None |
Exam Keyword Flashcards
Tap a card to flip it.
Quick Self-Check
Four questions pulled straight from the module.