Overview of Backup Technologies and Major Backup Software

Backup Evolution

Backup refers to copying all or part of data from a production host to another storage medium to prevent data loss caused by human error or system failure. Over time, several networking models have emerged: Host, LAN, LAN‑free, Server‑free, and Server‑less.

Host Backup

Host backup is the earliest method, where data is copied directly from the production server to backup media. It offers fast transfer speed and simple management, but lacks sharing capabilities and is unsuitable for large‑scale backup requirements.

Comment: Host backup is similar to copying data from C: to D:; its speed is high, but it cannot be shared across systems.

LAN Backup

In LAN backup, every production system installs a backup client that connects to a backup server over the network. The server controls the backup process and transfers data to the backup media.

Comment: LAN backup solves the sharing issue of Host backup but heavily depends on network and server resources, leading to possible bottlenecks.

LAN‑free Backup

LAN‑free introduces a backup client on the production system and a media management software. The backup server only sends commands; the production system directly transfers data to the backup media.

Comment: LAN‑free frees network resources and reduces server load, yet the production system still bears significant load.

Server‑free Backup

Server‑free combines snapshot technology with backup. A snapshot of the production data is created, presented to the backup server, which then mounts the snapshot and copies its contents to the backup media. The backup window is essentially zero, imposing no load on the production system.

Comment: Server‑free eliminates server pressure by using instantaneous snapshots.

Server‑less Backup

Server‑less transfers data directly between the source storage and backup media (often between a storage system and a tape library) without caching on a server. Commands are issued to the source storage, resulting in high performance due to the storage system’s superior resources.

Two implementation methods are described: using SCSI‑3 Extend‑Copy commands and using the NDMP protocol.

Comment (SCSI‑3): Requires high‑end storage, has poor compatibility, and only supports volume‑level backup, so it is rarely used.

Comment (NDMP): NDMP is an industry‑standard TCP/IP protocol for NAS backup, widely adopted.

Mainstream Backup Software

EMC offers NetWorker (software‑only) and Avamar (appliance or virtual edition). Symantec provides NetBackup and Backup Exec. Commvault’s product is Simpana, and IBM’s solution is Tivoli Storage Manager (now IBM Spectrum Protect).

EMC NetWorker

NetWorker handles backup and archiving, supports NAS NDMP, and routes data from client to storage server or node, which then writes to various media (disk, tape, network share).

EMC Avamar

Acquired in 2006, Avamar focuses on data deduplication and is often paired with NetWorker; it offers both an appliance and a virtual edition (AVE) that can use third‑party storage.

Commvault Simpana

Simpana provides data protection, archiving, replication, resource management, and search modules, organized into three tiers: CommServe (control), Media Agent, and iDataAgent (client).

Symantec Backup Exec (BE)

Targeted at Windows environments, BE is a low‑end product supporting file backup and LAN‑free; it handles Windows and Linux clients but limited database support.

Symantec NetBackup (NBU)

NBU is a high‑end, cross‑platform solution supporting a wide range of operating systems and databases, offering both backup and archiving, and supporting LAN‑free and SAN‑free modes.

IBM Spectrum Protect (TSM)

TSM delivers enterprise‑grade data management, including backup, archiving, space management, and disaster recovery, with an agent‑server architecture similar to BE.

Backup Software Feature Analysis

Backup vs. Archiving

Backup stores data for recovery and retains multiple versions; archiving moves infrequently accessed data to cheaper media for long‑term retention.

NetWorker, Simpana, NBU, and TSM include both backup and archiving; BE requires an additional product for archiving.

Data Deduplication

Deduplication reduces redundant data blocks. NetWorker lacks built‑in deduplication (requires Avamar or Data Domain), while Simpana, BE, NBU, and TSM have native deduplication.

NDMP Support

NetWorker, Simpana, BE, NBU, and TSM all support NDMP, enabling generic NAS backup without vendor lock‑in.

OS Compatibility

TSM, Simpana, and NBU offer comprehensive OS support; NetWorker is moderate; BE has the least coverage.

Maintainability

TSM’s installation is complex, whereas other products are easier to manage. Overall, functional differences are minor; pricing and target customers vary.

NAS Backup and NDMP Technology

NDMP originated from NetApp and Legato (2003) and has evolved to version 5, with version 4 being widely adopted. Most NAS vendors now support NDMP.

Traditional NAS Backup

Traditional methods read data from NAS via CIFS/NFS and transfer it over the network to a backup server, consuming network and server resources.

Comment: This approach suffers from low performance and high resource usage for large data volumes.

NDMP Functions

NDMP allows NAS devices to send data directly to attached tape libraries or backup servers without a client agent. It supports 2‑way and 3‑way networking.

2‑Way Networking

The backup medium connects directly to the NAS; the backup server issues commands, and the NAS writes data straight to the tape library, improving performance but limiting the tape library to a single NAS.

3‑Way Networking

Data flows from the NAS to a backup server over a dedicated network, then to a tape library, allowing multiple NAS devices to share the tape library at the cost of network bandwidth.

Remote Networking

The backup software handles tape operations while the NAS only sends data to the backup server, reducing NAS requirements and enabling tape sharing, though performance may be lower due to network transfer.

These three NDMP topologies each have trade‑offs and should be chosen based on specific scenario requirements.