The GOOD Way to Back Up Your Hard Drive

by Jeffrey A. Lomicka Tidbit Software Engineering Company

As computer users, we are often advised to back up our data, but rarely advised as to how. With this article and the GOOD Backup Utility, I hope to answer that question.

Why make backups?

Information is valuable, and time is valuable. If you store any original work on your computer, software, music, manuscripts, artwork, financial records, no matter what it is, you are one mousestroke away from losing it forever. The most unreliable component in your computer system is the computer user. Most of the time, when I need to restore a file, it's because I did something stupid. If we could get all the bugs out of that computer we keep between our ears, we could get by with fewer backups. Computer hardware is failable. I have a letter from a GOOD Backup Utility user whose disk power supply died right after doing a backup. Computer software is fallible. The bugs and limitations of the software we use, such as the GEMDOS fourty folder limit is a constant risk. I have also known people to have their entire computer system and hard drive stolen. While rare, this does happen.

What, then, is a GOOD backup?

A good backup must address our basic assumption that information is valuable, and time is valuable. This means that you have to do backups frequently, or you run the risk of losing valuable information, and therefore, a backup can't take much time to do. If it takes too long to do a backup, you won't do it very often, increasing your exposure to unrecoverable loss of information. In addition to running quickly, the backup must give you confidence that you can actually get files back if you have to. If you don't have confidence that the backup is useful, you won't be motivated to keep it up to date, or you may be unable to actually do the file restore when the time comes that you need to.

GOOD Backup Utility Design Goals:

After attending to the required mechanical details, the features of the GOOD Backup Utility are all intended to support one of two major goals. These goals are to reduce the time needed to keep backups that are always up to date, and to increase the confidence that files on both the active disk drive, and the back disks, are complete, intact, and recoverable.

Choice of backup media:

The trade-off is mostly between cost an speed. Other concerns are total capacity, and flexibility with respect to recovery procedures and updates.

Floppy disks win big on cost, with a low initial investment (who doesn't already have a drive?) and with media costing at well under $1.00/megabyte. There is an unlimited total capacity, allowing you to keep multiple backups, and to periodically archive your data. Floppies are also very flexible, but are fairly slow, and require attention during the backup to swap in and out, although with the incremental update procedure of the GOOD Backup Utility, floppies can be an excellent choice even for large capacity drives.

A second hard drive is fast alternative, and their capacity allows the backup to run unattended, but at a cost of about $10/megabyte, it can be difficult to justify an extra drive just for a backup.

Streaming tape drives are best suited for Image backups. Although they can support some very large capacities, and large data transfer rates, these advantages must be weighed against the need to scan the tape sequentially to locate a single file, and the bookeeping problem of trying to organize an incremental backup system, since you can't update them in place. Tape drives score very low for me in the confidence factor. I like to be able to put the backup disk in my system and say, Look, there are my files! The GOOD Backup Utility doesn't support tape drives, or do image backups.

Removable Cartridge Drives are perhaps the best backup media. They have all the advantages of floppy disks, in terms of unlimited capacity, flexibility, and confidence, and they also have all of the advantages of hard drives, in terms of speed and unattended operation. The media cost is only about twice that of floppies, around $2.00/MB. Their primary disadvantage is the $700-$800 initial cost of the drive.

How often is often enough?

Never wait so long that you would be unwilling to lose what's there. For most ST users, backups should be done weekly, and always immediately after a creative or time-consuming session with the computer. In a professional environment, backups should be done every day, as a matter of routine.

How Many Backups should I keep?

I recommend that you keep two sets of backup disks (savesets), store one well away from your computer, and swap them each week. This is particularly true if you use floppy disks. I have found the floppy disks are not the worlds most reliable way of storing data. This also protects you from failures while backup is in progress, and from annoying events like having your backup disks stolen at the same time as your hard drive.

Note that the desire to keep and update two separate backups of the same hard drive means that the notorious Archive Bit is completely useless. The GOOD Backup Utility will back up and restore the archive bit intact, but otherwise it ignores it completely. The archive status is stored in a separate data file that is stored with the backup, not as a tag on each file.

Are incremental backup systems good?

Incremental backups are good because when you perform a backup, you only have to copy the files that have actually changed since the previous backup. Several days of hard work will usually have only a few files that have changed, so the backup always goes quickly. There are two incremental backup methods. One is good, and the other is not so good.

A traditional incremental backup, based on ancient tape based systems, is one where you start with a complete save of everything. Then, at a later time, you save only what changed onto an additional disk or tape. The collection of backup disks (or tapes) grows and grows until you decide to do another time consuming complete save. These systems rarely keep track of what is deleted, which means if you have to do a restore, any work you did in deciding what to delete is lost work.

A good incremental backup is also started by saving everything once. Then, at a later time, you delete out of the backup any files that are now obsolete, by having been deleted or modified, and then re-use that space to copy over any files that are new or modified. The result is a complete backup of the hard drive, with nothing missing, and nothing extra, and with all the time saving advantages of incremental operation.

What about Raw Copying Speed?

The slowest part of disk input/output is stepping the disk heads from track to track. These are the grinding and kerchunking sounds made by your floppy disk drive. Disk I/O is much faster if these tracks are read and written in sequential order, rather than in random order. The GOOD Backup Utility accomplishes this by use of a write-back, look ahead, most recently used disk cache. The operation of the cache insures that most of the time, disk I/O is performed track by track in sequential order. To accomodate users that have extended memory available, the size of this cache is user configurable. The more cache memory you give it, up to the size of the disk itself, the faster it goes.

Never be a slave to your machine.

The GOOD Backup Utility can be interrupted at any time. Just hit any key on the keyboard, and at the end of the current file copy, it asks Continue or Quit?. If you choose quit, you can stop right there, and finish the backup some other time.

Also, if I run out of disks, I can stop until I get more, without sacrificing any work I have already done.

How Can I have Confidence in the Backup?

Start by using a utility whose backup disks can be read directly by TOS. If you ever want to be sure the files are there, you can read then right off the backup disks.

In the GOOD Backup Utility, as each file is saved, a checksum is computed and stored. This checksum is computed from the actual contents of the file, and can be used to verify the quality of both the backup disks and the hard disk. On each update of the backup, the old files that have not been modified are all verified. This detects any serriptitious modifications that might have happened between backups.

This is one of the things that makes the GOOD Backup Utility better than the traditional techniques. With traditional image backups, you could easily reuse a set of backup disks several times before even noticing that a file was corrupted. By then, you might have wiped out all available backup copies. GOOD will detect this before it is too late.

The checksum values can also be used to verify the backup disks. This is a task you might want to do if you intend to reformat a disk or move to another drive.

What about things that just make it easier to use?

Context sensitive help explains every step. This reduces time spent finding things in the manual. Any backup disk can be replaced without redoing the whole backup. This save times in the event a disk is lost or damaged. Groups of files can be included or excluded from the backup the backup using wildcards. Floppy disks are formatted as you need them. A unique incremental restore capability means you don't have to restore a backup all at once.

Now available as shareware

The shareware fee is $15, by cash or check to "Jeffrey Lomicka", 25 Wood Lane, Maynard, MA 01754.

Go to Jeffrey Lomicka's page
Comments to Jeff: lomicka at mva.net