Backing Up Your Data
Balance expense and effort with risk
May 14, 2002
As the conservation movement's use of computers increases, so does the importance of making regular backups of your key data. How effective would you be if your email, word processing documents and contact database were wiped out? How many hours would it take to rebuild that information from scratch?
Backing up your data regularly is vital insurance against a "data catastrophe." Unfortunately, this is a lesson that most people learn only from bitter experience. Developing a solid backup plan requires some investment of time and money, but the cost is far less than the often-impossible task of recreating data for which no backup exists!
Backup Best Practices
All backup routines must balance expense and effort against risk. Few backup routines 100% airtight -- and those that are may be more trouble to implement than they're worth. That said, here are a couple of rules of thumb that should guide you in developing a backup strategy:
- Develop a written backup plan that tells you:
- What's being backed up
- Where it's being backed up to
- How often backups are occurring
- Who's in charge of performing backups
- Who's in charge of monitoring whether backups are being performed successfully
- Your database and accounting files are your most critical data assets. They should be backed up before and after any significant amount of data entry and/or use. For most groups, this means backing these files up every day. Groups that do a lot of data entry should consider backing up their database after each major data entry session.
- You should back up your other documents (e.g. "My Documents" folders) and email files at least once a week, and probably once per day. Each organization needs to decide how much work they're willing to risk losing, and set a frequency of backups accordingly.
- You should store a copy of your backups off-site to insure against a site-specific disaster such as a fire, break-in or flood. Ideally, you should store your backups in a safety deposit box. Generally, we recommend rotating a set of backups off-site once per week.
- It is not usually necessary to back up the complete contents of each hard drive -- most of that space is taken up the operating system and program files, which can be easily reloaded from CD if necessary. The only exception is if your organization has a dedicated file server; it's a good practice to do a full backup of your server so that have a way to restore your server's entire hard drive.
- Don't forget to back up your laptops! Laptops can be hard to back up, but are just as important as your desktop workstations.
- Test your backups BEFORE you need them. You need confidence in your backups. Make sure your backup software has full read-back verification. Try restoring a few files.
Choosing Backup Hardware
Choosing appropriate backup hardware is the first key to implementing an effective backup strategy. In any situation, there are probably several "right" answers. Here are some guidelines for choosing backup hardware that will work for you:
- Determine how much data you need to back up. Take a look at each machine on your network -- or at least a representative sample. How big is the each user's documents folder? How big is the email file? How much data do you in your organization's primary shared folder? Add up the totals for all your machines, or multiply the average by the number of machines in your organization. Be sure to leave room to add a few new staffers, and to plan for growth -- it's not impossible to add 100 MB of email per person per year.
-
Choose a backup device that uses tapes or cartridges with a capacity that's at least twice the total amount of data you need to back up. This will give you room for growth, and will also allow you to perform "incremental" backups on the same tape with a "full" backup. (More on this in a bit.)
For many organizations, DAT DDS-4 tape drives are a great choice, as these drives combine excellent reliability and reasonably fast speeds with large storage capacities (20-40 GB per tape) and reasonable media costs (~$25 per tape). DAT DDS-4 drives cost around $800.
Travan TR-5 drives are a bit less expensive (around $300). Their tapes hold 10-20 GB of data, and cost around $40 each. Travan drives are slower than DAT drives, and considered to be slightly less reliable. However, their low cost and solid performance make them a good choice for many environmental organizations.
Both DAT and Travan drives are widely-used standards and hardware is made by a variety of companies.
The third reasonable choice for mid-size networks is the Onstream Echo 30GB ADR drive. This drive costs about $250 and uses 30 GB cartridges that cost around $50 each. It's a proprietary format from a relatively new company. In today's world of fast-disappearing technology companies, there is some risk that the Echo format could be "orphaned" with little notice, but the Echo's large capacity and low price are compelling to small organizations with big data storage needs.
- Another consideration is the speed of the drive and its interface to your computer. If you have a large amount of data to back up, having a big storage device isn't much good if you can't write data to it quickly. There are three common ways to connect a backup drive to your machine: SCSI, IDE and FireWire SCSI drives, which can be internal or external, are the fastest -- and most expensive. Most PC's don't have built-in SCSI adapters, so you may need to add a SCSI card ($50-$150). (Some servers have built-in SCSI, though.) IDE is used for internal drives only, and is inexpensive, with middling performance. All PCs have built-in IDE connections. Firewire is a new format for connecting external drives. Firewire devices are mid-priced, and offer mid-range speed. New Macs have built-in Firewire ports; PCs need to have Firewire cards added (~$100).
What About CD-RW and Zip Drives?
CD-RW and Zip drives are inexpensive and lots of folks now have them. Why not use them as your primary backup device? Here are a few considerations:
- Limited capacity:
- CD-RW discs only hold about 650 MB of data per disc; that's plenty to back up your database, or a single person's documents and email, but it's not enough space to hold the files of everyone in a 5-20 person organization. Zip disks are even smaller, at 100-250 MB per disk.
- Finicky hardware:
- CD-RW drives can be finicky, and sometimes produce "coasters" -- failed recordings. Not something you want to base a backup strategy on. Zip disks are relatively fragile. And external Zip drives (especially older parallel port Zip drives) can be VERY slow.
Does that mean that CD-RW and Zip drives are useless? Definitely not! Here's how we think you should use CD-RW and Zip drives:
- Use CDs for archiving old data:
- CD-R disks are very cheap: about $1 for each 650 MB disk. They're durable. And they can be read in any machine with a CD-ROM drive. These qualities make them uniquely well suited for archiving data that you don't need to change after it's been created. Photos and finished print publications are two great examples. Both generally involve large files that you need to keep around, but you're not likely to go back and change. Archiving old data files to CD-R is a great supplement to a tape-based backup strategy, because it lets you avoid wasting resources backing up big chunks of files that won't change. Plus, it makes your archives portable -- and easy to store a copy off-site.
- Use Zip disks for transferring files, or as a secondary backup:
- Zip disks are great for making quick, easy redundant backups of super-critical files such as databases and accounting files. We think it's a great idea to use a Zip drive on your accounting and/or database machines to make an extra backup of your most critical files.
Backup Hardware Compared
The table below summarizes some of the leading choices for backup hardware. Use it with the advice above to help you choose backup hardware that's appropriate for your organization.
| Device | Hardware Cost | Media Type/Capacity | Media Cost | Notes |
|---|---|---|---|---|
| Iomega Zip Drive |
250 MB 100 MB |
Magnetic disk / 250 MB Magnetic disk / 100 MB |
$20 $15 |
Recommend USB, SCSI or internal IDE models for better performance; parallel port drives are very slow. |
| Castlewood Orb 2.2 GB drive | $150 internal EIDE $160 internal SCSI $180 external SCSI $200 USB |
3.5" Cartridge | $30 | Great price for storage capacity and speed. Reliability has been mixed, but much improved as of late. |
| Travan TR-5 format tape drive |
$250 - 600 internal or external, SCSI or parallel models |
Tape cartridge / 10 -20 GB | $40 | Popular tape format. Slower than DAT, but inexpensive. Large capacity per tape. |
| OnStream ECHO ADR format tape drive |
$150 - 600 internal IDE, internal / external SCSI, external USB, external parallel models |
ADR tape cartridge / 15 - 30 GB (25 - 50 GB format drives / tapes also available) |
$45 |
Proprietary format. Inexpensive high capacity, good performance. Internal models preferred. |
| DAT DDS-3 format tape drive | $500 - 800 internal or external SCSI |
4mm DAT Tape / 12/24 GB |
<$20 | Midrange DAT format. Fast for tape, very cheap media. Costly hardware. |
| DAT DDS-4 format tape drive | $800 - 1100 internal or external SCSI |
4mm DAT Tape / 20/40 GB |
<$20 | High-end DAT format. Relatively fast, affordable media. |
| DLT format tape drive | $1200 - $1500 internal or external SCSI |
DLT IV tape (40/80 GB) | ~$80 | High-capacity tape format; can scale to very large auto-load units. |
Sample Backup Scenarios
Here are three sample backup scenarios appropriate to single or several standalone users, a small (3-7 person) network, and a larger office network.
Standalone user(s)
- Who:
- A single user, or several people who are in the same office, but do not have a local area network ( LAN) connecting their machines.
- Hardware:
-
The Iomega Zip drive is an inexpensive (<$200), easy to use drive that uses 100-250 MB disks that resemble floppies. These cartridges are fairly inexpensive and versatile. The Zip has many versions designed to attach to a PC or Mac any number of ways. The advantages of the Zip drive are that it is easy to attach to several different computers (especially the USB version), uses relatively inexpensive media, and holds enough data to easily back up most people's personal files. The SCSI, USB and internal IDE models are fairly quick, the external parallel port models quite a bit slower.
Something else to consider is the fact that the Zip has been around for a while and has become something of standard in the low-end removable-media market. Many print service bureaus use Zip media to transfer/receive large desktop publishing files. If you need to exchange files with other people this way, the Zip might be a good choice. On the other hand, Iomega has garnered a reputation for low-quality customer support, and some people have experienced quality problems with Zip drives resulting in damaged disks and lost data.
A single user with a large amount of critical data to back up (such as someone doing intensive database or GIS work) might consider a OnStream Echo or Travan TR-5 drive, which have a multi-gigabyte storage capacity.
- Software:
- Macintosh:
- We recommend that you store all of your personal files in a single master folder and back that entire folder up each week by simply copying the file to your Zip drive. If you'd prefer to automate the process, or want the added security of backing up your entire hard disk, we recommend Retrospect Express ($50), from Dantz. Retrospect Express is a simple, inexpensive and powerful backup program that should meet all the needs of a single user or small office network. Synk X is an easy-to-use shareware product that can back up to most mountable drives.
- PC:
- We recommend using Retrospect Express ($60). Retrospect Express is a powerful and easy to use backup program that will help you manage your backups quite effectively. Some backup drives also ship with adequate backup software. For example, Iomega Backup, which ships with all Zip and Jaz drives, is much improved of late, and is freely downloadable from the Iomega Web site. The OnStream Echo drive also ships with serviceable software for a standalone user.
Small office
- Who:
- A 3-7 person office with computers that are connected by an Ethernet LAN.
- Hardware:
-
A Travan TR-5 format tape drive such as the Seagate TapeStor series of drives. These tape drives are relatively inexpensive and offer large storage capacities (up to 20GB on a single TR-5 tape). The OnStream ECHO drives are also a reasonable choice, with a 30GB per tape capacity.
Networks that need to store many sets of backup data might choose to use DAT DDS-3 or DDS-4 drive. The hardware cost is a bit higher than for Travan drives, but the media is less expensive, longer lasting and faster.
A high-capacity removable cartridge drive such as the Castlewood Orb would also be a reasonable choice for organizations that frequently need immediate access to their data archives. Orb drives use 2.2 GB removable cartridges with capacities Cartridges cost about $40 each--rather expensive compared to the much larger $20-30 Travan or DAT tapes. Orb drives (particularly the SCSI and internal IDE models) are very fast, and thus more convenient than tape to access frequently, and can store much more information per cartridge than the Zip drive. However, they cannot compete with the high storage capacity or low media cost of tape backups.
In addition, small offices may also consider purchasing a CD-RW drive. It can supplement your primary backup system by providing a means to inexpensively archive old data. CDs are particularly well-suited for archiving data that doesn't change after it's created, such as newsletters and photos. CDs offer very low cost-per-megabyte archival storage, and CDs are both durable and easy to use..
- Software:
- Macintosh:
- Retrospect Workgroup Backup ($300) is an appropriate choice for backing up a network of machines.
- PC:
- Retrospect Workgroup Backup ($300) or Veritas Backup Exec Desktop Pro ($130) are reasonable choices. Retrospect offers more advanced functionality, but Veritas Backup Exec Desktop Pro works almost as well in a peer-to-peer environment, but doesn't support NT/2000 servers.
Larger office
- Who:
- Offices with more than seven workstations, or with a Windows NT/2000 server, connected via an Ethernet LAN
- Hardware:
-
Tape drives are the backup media of choice for networks larger than seven workstations. Although tape drives are slower than removable-cartridge hard drives, their media (tapes) is far cheaper than removable hard disk cartridges, making them the only affordable solution that allows multi-gigabyte backups. The two affordable tape choices are Travan and DAT.
Travan TR-5 format tape drive. A typical Travan TR5 format drive costs about $350 and can store up to 8 GB of compressed data on a single tape. Tapes cost about $40 each.
DAT DDS-3 or DDS-4 format tape drives are more expensive to purchase, typically in the $500-800 range, but their media (4mm DAT) are extremely inexpensive ($5-7 for a 4 GB cartridge, $10-20 for a 12/24 GB tape), making DAT a very cost-effective choice for doing regular full backups of all machines in a network. DAT tapes have also proven to be a bit more sturdy--lasting longer than Travan tapes.
- Software:
- Macintosh:
- Retrospect Workgroup Backup ($300), which is sometimes bundled with some tape drives.
- PC:
- Retrospect Workgroup Backup ($300) is probably the best choice for a midsize office, as it offers highly manageable network backups. The latest versions can even backup an Exchange server. Veritas Backup Exec ($600) is higher-end software that has many features suitable for backing up Windows NT/2000 servers that are doing more than just basic file sharing.