Disaster Recovery Explained: Definitions, Strategies, and Implementation

1. Definition of Disaster Recovery

1.1 What is Disaster Recovery?

Disaster recovery (DR) refers to the use of existing scientific and technical means to establish reliable emergency procedures in advance to cope with sudden incidents.

DR includes both backup systems and disaster‑recovery systems.

1.2 Backup and Disaster‑Recovery Concepts

Backup

Backup: Ensuring data safety . It is the process of copying all or part of a data set from production disks or arrays to other storage media.

Disaster Recovery

Disaster Recovery: Ensuring business continuity . It builds one or more remote IT systems that contain complete infrastructure (compute, network, storage, power, cooling). When the primary data center fails, the secondary site can quickly restore services.

Differences

Protection objects: Backup protects data; disaster recovery protects business continuity.

Implementation: Backup uses backup software; disaster recovery uses replication or mirroring software.

Time window: Replication/mirroring has a much shorter protection cycle than backup.

2. Role of Disaster Recovery

2.1 Problems in Data Centers

Viruses, OS vulnerabilities

Human error

Terrorist attacks

Power failures

Hardware failures

Natural disasters (earthquake, flood, typhoon)

2.2 Consequences of No DR

Business interruption

Data loss

Customer complaints

Revenue decline

Compensation costs

Company bankruptcy

2.3 Backup Functions

Storage Layer

Backup consists of five parts: backup client, storage strategy (media, deduplication, retention, write I/O), backup content (what to back up, what to exclude), backup policy (deduplication, type, schedule), and performance optimization (client read streams).

Cloud Computing Layer

Cloud Server Backup Service (CSBS) provides whole‑machine backup using consistent snapshot technology and remote replication. Volume Backup Service (VBS) creates backups of cloud disks and can roll back data.

Replication Types

Synchronous replication copies data in real time; asynchronous replication copies data with possible consistency lag.

2.4 Disaster‑Recovery Scenarios

Local high‑availability (HA)

Active‑standby (AS)

Active‑active (AA) data centers

Two‑site three‑center (3DC) solutions

Local HA

Uses real‑time mirroring and synchronous replication within the same campus; typically targets RPO≈0.

Active‑Standby

Provides near‑zero RPO and low TCO; supports automated failover and recovery.

Active‑Active

Six‑layer active‑active architecture; aims for zero business interruption and zero data loss.

Two‑Site Three‑Center

Offers both cascade and parallel networking options, each with its own advantages and performance requirements.

3. Measuring Disaster Recovery

3.1 Backup Types

Full backup – complete copy of all data at a point in time.

Cumulative incremental backup – incremental changes since the last full backup.

Differential incremental backup – changes since the last backup of any type.

3.2 Backup Strategy Principles

Combine full backups with either cumulative or differential incremental backups, but avoid mixing both incremental methods in the same policy.

Choose the combination based on space and backup‑window constraints.

3.3 DR Metrics

Recovery Point Objective (RPO): Maximum tolerable data loss measured in time.

Recovery Time Objective (RTO): Maximum tolerable business downtime measured in time.

Standard levels (GB/T 20988‑2007) map RPO/RTO to recovery‑ability grades from 1 (days) to 6 (minutes).

4. Implementing Disaster Recovery

4.1 Backup Methods

LAN‑Base – backup agents on production servers send data over LAN to a backup server.

LAN‑Free – data bypasses LAN, flowing directly from file servers through FC switches to tape.

Server‑Free – data is copied directly from storage to tape without passing through the server’s CPU, memory, or bus.

NDMP – network data management protocol enables storage devices to send data directly to backup targets.

4.2 Backup Media

Disk arrays

Tape libraries

Virtual tape libraries

Optical libraries

Cloud storage

Integrated appliances (e.g., Huawei HDP3500E)

4.3 Design Principles

Customer requirements (data type, volume, objects).

Backup policy (frequency, schedule).

Network planning (bandwidth, topology).

Storage planning (capacity, growth).

4.4 Disaster‑Recovery Techniques

Host‑level replication – software installed on servers replicates data.

Network‑level replication – similar software uses network links for replication.

Storage‑level replication – dedicated storage systems replicate data over fiber or WAN, minimizing impact on servers.