Printer Friendly

Software approach to network backup.

Backing up data on a network can be a pain in the neck. The result is that less than half the data on most network is backed up.

The impact of this is compounded because the number of gigabytes on a typical network is growing by 50% a year, and terabyte-capacity networks already are in place. That's a lot of data to be parked on local disks, using up storage space and ready to disappear forever at the first crash.

What's worse is money for the capital equipment and people necessary to manage backup is scarce.

Storing all the programs and data on a file server and depending on a local department administrator to dump it onto tape every night is one answer. But how about a strategy that saves money by using backup resources already in place and takes user and network requirements into consideration when doing the backup?

The network administrator won't have to run backups during working hours, which eliminates user complaints about downtime and poor response time. It also means the network administrator doesn't have to hang around at night to run the backups. That can be important if you're the network administrator.

One way to solve backup problems is to implement a decentralized, software-based backup approach. This is especially appropriate in Unix-based networks where the native backup utilities are limited.

For example, dd's complex syntax is hard to learn and use; volcopy, while fast, doesn't give you a lot of flexibility; and cpio and tar are slow. (The standard Unix backup utilities also do not have a volume labeling mechanism, so there is no way for an organization to know on which volume a specific file resides or prevent a volume from being overwritten.)

Backup solutions have to be economical, able to adapt to current and future needs, and contribute to network efficiency. Off-the-shelf network backup software, implemented on existing hardware, is the approach that best meets these needs.


The primary benefit of a software approach is economy. The economy stems from three factors.

First, a user can integrate backup hardware already in place. One organization with several thousand workstations had servers in place they wanted to back up.

They had a lot of tape drives in place, and didn't want to add more if they could avoid it. So they bought a software backup solution that made their existing tape drives a network backup resource.

Using existing hardware cuts the cost to get started with integrated backup and keeps the price-per-node of backup low. It is an important factor in today's heterogeneous networks, wher systems and tape drives from many manufacturers are tied together.

A full-functioned software package can use equipment from every manufacturer in the network. For example, a user on a Sun workstation can back up files to a tape drive on a Digital Equipment Corp. VAX computer.

The second contributor to economy is the ability to share distributed taped drives. Of course, there's nothing new about sharing tape drives; Unix users have been doing that with NFS and standard utilities for years.

But these utilities don't let the drives be shared intelligently, because they have no volume management and tape drive device management. That gives a network administrator no way to determine where a file has been stored.

Detailed volume labeling and cataloging, volume management, and an on-line catalog let an administrator go straight to the correct tape and restore the file, no matter what volume the file is on and where in the network the volume is located.

That's a solution to every network administrator's worst dream: Being asked to restore a file that existed once, but was lost some other time--no one knows when. It would be an impossible task in most organizations to find and restore that file.

Backup software accepts the request, identifies the file, and tells the administrator which tape to mount.

Third, backup software lets users do more with fewer drives. By making each drive a network resource, more users have access to it and it is used more efficiently. That's one less drive to be purchased.


A software backup strategy enables administrators to define the roles of the backup drives more flexibly than they can with any other kind of strategy. So an administrator can, for example, have one tape drive for each server in an engineering department.

Or, she can share one drive over many servers in one area. Or form a centralized media pool by equipping one server with many tapes to back up applications with massive files, or back up those critical production applications every half hour.

A software approach also provides redundancy. If one drive in a multiple-drive network fails, the software will automatically bypass the failed drive and route requests to one of the remaining drives.


A distributed backup approach can help increase the efficiency of the network.

If an administrator's network has hundreds of gigabytes on each of several servers, she usually does not run out of backup storage capacity; she runs out of time to back up all these files. devices aren't fast enough, network bandwidth isn't wide enough, and days aren't long enough for the administrator to back up a large network through a central point.

This problem will only get worse as networks expand to terabyte capacities.

A distributed backup approach can solve this problem by localizing backups and restores at the workgroup level. Multiple simultaneous backups can run inside a large network's partitions to maximize the amount of data transferred to tape, while minimizing the amount of traffice placed on the backbone network.

Another way to increase efficiency is to program the server to back up parts of the network at different times of the day.

This isn't as complex as it seems. If the network has enough bandwidth to back up all these files, the administrator can use the Unix cron command to back it up at night. Cron starts a process at predetermined times.

Another approach is to use worklists, which can be generated from menus or command lines in a few seconds and run during slow times on the network.

Worklists let an administrator control what's backed up. So instead of backing up all the files, an administrator can be selective, always taking password files, for example, and bypassing temporary files. This has the effect of speeding up backups by ignoring unwanted data.

Efficiency can be increased by backing up users while they are online, so they can stay productive. This requires a file locking facility. Unix systems have file locking capability used by backup and restore software.

These alternatives provide users with ways to back up files, and help reduce the amount of backup traffic on the network, keeping central server bandwidth available for online work.

Hardware-based solutions

There are more ways than a distributed software approach to implement backup strategies. Administrators also can buy an integrated server that includes hardware and software, or buy storage hardware and write their own software.

Integrated servers typically incorporate a storage hierarchy made up of magnetic disks for online storage, tape or erasable optical disks for backup, and WORM (Write Once Read Many) optical disks for archiving. The hardware and software were designed to work together. If a user wants a centralized backup facility, these systems fit in well.

A centralized integrated system works well for some applications, typically those with large files. Their hierarchical approach means that when users access a file on a server, they are held in magnetic storage.

When they have to be backed up, the server copies the data to its own on-board tape or erasable optical disk drives. Integrated servers can back up themselves fairly efficiently, but cannot back up the local files on a workstation.

Suppliers of integrated servers also provide software for their systems that eliminates the need for users to do low-level coding. The software can range from drivers and programming tools to data management facilities with library and volume management, and backup and recovery routines.

There is no "best" backup strategy, because the needs of organizations are so different. Every administration should seek an approach that takes advantage of equipment already installed to keep the network uncluttered, provides for the inevitable increase in workstations and traffic, and makes backup less of a bother for users.
COPYRIGHT 1991 Nelson Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Brooks, Paula
Publication:Communications News
Date:Jun 1, 1991
Previous Article:Resource management: how two users manage the known.
Next Article:Prefabricated modular buildings: lower cost, fewer headaches.

Related Articles
Maxtor Goes To The Max With MaxAttach And Reflect-It.
Solve the data-protection-and-recovery puzzle.
NAS/Tape Backup Strategies Cope With New Data Cornucopia.
Breakthroughs In Enterprise Backup Solutions For NAS File Servers.
Recovery for the rest of us: small and mid-size businesses need to be as diligent as large enterprises in implementing an emergency preparedness...
SAN-based intelligence: the Holy Grail of storage management?
Hardware enables next-gen storage: next generation storage software will require specialized hardware.
Policies, precautions and practices: some facts and practices for disaster recovery.
The push for continuous data protection.
Watching your back (up): prevent your data from disappearing.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters