Backup Made Easy

Backing-up one’s work has always been a bit of a problem for single PC users, and has become a much greater problem as disk drives have increased in size so enormously, and data files have tended to become commonly very large. At one time it was practical to rely on floppy disks but not any longer in most cases. Today the wide availability of CD-R has changed the whole scene.

History of Backups

Users on office networks generally have the matter taken care of by the administrator of the server computer, who runs mass backups every night on high speed, high capacity tape drives. Such tape drives (such as DAT) is very efficient, reliable and fast, but also rather costly. For small groups or single users, a common medium has been the low cost mini tape drives such as Colorado and Travan. However these tape systems were never completely satisfactory because of their lack of speed and fact that reliability left something to be desired.

The advent of the Iomega ZIP drive a few years back, and its rapid and widespread acceptance, looked like the answer to most small users backup needs. But it didn’t take long for its 100Mb capacity to become inadequate for many users as file sizes continued to bloat. The 1Gb and 2Gb Jazz drives never gained such wide usage because of their high drive and media cost.

In the last year or so, the availability of recordable CD drives has moved from very costly specialist items to very low cost that makes them usable as an alternative to the CD-ROM drive. Many new PCs, even in the domestic price range, are now being sold with a CD-RW drive instead of a CD-ROM drive. This, together with the extremely low cost of the media, revolutionises the backup scene.

That leaves the matter of how to organise your backup technique. As well as deciding on an operational strategy, there are some technical options to choose between in creating CDs, as well as two different media types – write-once CD-R and re-writeable CD-RW.

As the name implies, CD-R disks can only be written to once, and not erased and re-recorded. However, the CD-writing software such as Adaptec’s EZ-CD Creator provide facilities which make it appear that the CD-R disks can be erased and altered. But what happens is that the current data is locked-out in the file system directory table, and if it is being altered, the altered data is written in a new location on the disk and the directory table changed to link the data blocks accordingly. This gives the effect of erasing and/or updating data, but each time data is changed more disk space is used up and also more directory space, because the same alteration method has to be used for the table as well. If changes are made too often this can lead to running out of disk space unexpectedly. So it is best to avoid changes to data on CD-Rs except occasionally, and use CD-RW for data that is likely to be updated.

Data on CD-RW disks can be erased and updated as often as desired, in-situ, just like a magnetic disk. The disks are significantly dearer but still not expensive. The major limitation compared to CD-R is that a CD-RW disk can only be read in a CD-RW drive, whereas a CD-R disk, if written using a suitable option, can be read in any ordinary CD-ROM drive, except for some very old ones. So CD-RW is best for regular daily working backup where it will only ever be used in the same CD-RW drive. It can be accessed just like a big floppy disk or ZIP disk.

CD-R is ideal for cumulative archive backups. The very cheap media ($1 to $2 for 650Mb) allows frequent creation of new disks without major cost consideration.

CD-R disks can be created using two main recording techniques – continuous track recording and packet-mode recording. Continuous track recording requires you to use the special CD-R software and its internal facilities for selecting the data on your system that is to be recorded. Then it is all recorded in one fairly long operation. The result is a disk that can be read in any standard CD-ROM drive, unless some problem interrupted the recording stream enough to break the continuity. In that case you’ve made a dud CD that will have to be thrown away. That used to be a common problem, but the latest CD-R software manages the data flow much better, helped by the higher speed of the newer PCs.

The Packet-Mode method sets up a directory system on the disk and records data in fixed-length ‘packets’ somewhat analogous to the sectors of a floppy or hard disk. This enables you to place data on the CD gradually in small lots such as a file at a time, just like writing to a floppy disk. The CD interface software sets up the CD’s data structure, which takes a few minutes, and then exits but stays running in the background. From then on you can write to it using your ordinary file manager tools such as Windows Explorer, My Computer etc. The CD-R disk appears just like a CD-RW disk, but as explained, any erase-rewrite operations lock out disk space.

When you eject a CD-R that was set up for packet mode writing, the CD-R writing software pops up and asks if you want to leave it in that mode so that more data can be written later, or converted to normal CD-ROM mode, so that it can be read in an ordinary drive. If left in packet-mode it can have extra data written to it later, but can only be read in a CD-R or CD-RW drive controlled by similar packet-mode CD-R interface software. If you choose the ‘convert’ option, the data indexing on the disk is re-organised, and then it can be read in an ordinary drive. It is still possible to write more data to it later in most cases, but that requires the disk to be made into a multi-session disk or converted again into packet-mode. Each packet-mode conversion consumes more disk space, and each additional ‘session’ in the multi-session method consumes quite a bit of disk space, so generally it is best to avoid multiple conversions.

I will describe my own backup strategy to serve as a suggestion. Each person needs to figure out a strategy appropriate to their particular type of work and circumstances. My scheme is a multi-layered system to cater for immediate working data security, longer-term recovery of all data, the ability to quickly restore normal PC operation, and permanent archiving of work.

I work month-by-month mostly. I write material for monthly magazines, all of which have deadlines around the middle of each month. They send me software to install and write about, or ask me to download it from web sites. After I have written the story I may have to return the software, or it may be a time-limited version that stops working after, say, 30 days. Sometimes I get a full version and am allowed to keep it, but I wouldn’t want to keep every one of those items permanently on my system. (A few of them I’m glad to be rid of!) So in most cases I need to un-install the program. The un-install facilities in Windows are now usually quite effective fortunately, but this constant un-installing does tend to lead to problems eventually, mainly in the Registry, so I need to occasionally wipe and re-install the whole system.

I also need to remove the files I produced for each magazine story or job – text documents, scans and graphic screen captures – or else they would accumulate alarmingly. I typically generate about 50 megabytes of material for each story. Each 1280x1024 24-bit screen capture gives a 5Mb raster image. Scans of boxes and book covers and photographs are also usually around 5Mb each to give good enough sharpness for a glossy magazine, and I usually generate more of them than will end up being used. But I need to keep all the material for each job more or less permanently in case of later enquiries, re-working or even litigation. For work to be web-published, I have to produce low-resolution images, but I make those by reprocessing the original high-quality images and need to keep the originals as well.

I have a basic Windows installation and kit of working software – what large network administrators would call a ‘standard operating environment’. This comprises Microsoft Office 2000 Professional (Word, Excel, Access and PowerPoint), Ulead Photoimpact and associated raster image tools, HP scanner software, Adaptec EZ CD Creator & HP CD-RW interface, Iomega ZIP tools, Norton SystemWorks, WinZip, Internet tools and web authoring software, GlobalWriter multi-language word processor, Microsoft Visual Basic 6 and MSDN, Microsoft HTML and Windows Help authoring kits, Adobe Pagemaker, Acrobat Distiller & Reader, Zone Alarm firewall, Vet anti-virus, Corel Draw, Notetab text editor, MYOB accounting, and various update patches, printer drivers and extra fonts, plus my permanent CAD tools; AutoCAD LT 2000i, AutoCAD 2000i-Architectural Desktop, IntelliCAD 2000, Cadcorp SIS, and Piranesi.

That all comes to about 5Gb, and I installed them on an empty Windows 98se system created by the HP restore CD. Then I tweaked various Windows settings such as swap space, adjusted settings for the firewall, email operations, and default new data file settings, etc, and backed it all up to multiple CD-R’s using the Adaptec EZ-CD ‘Take-2’ backup program. Then I can quickly reinstate my basic system if a bad program mucks it up, or the system becomes ‘tired’ after repeated program un-installs has made the Registry a mess. I use the HP system recovery CD supplied with the PC, followed by restoring the ‘Take-2’ CD backup. I intend to use Norton Ghost to make a complete restore image, but so far I have failed to figure out how to do that.

I have a single hard drive of 10Gb, all as C drive with FAT32. I keep all my working data in a big folder tree called ’_Documents’, with numerous sub-folders. I also have a folder for downloads, and another called ‘_Info’ that I use as a sort of categorised cumulative information store. The ‘_Documents’ tree has sub-trees for business records, accounts, e-mail, personal files, user groups, and each client or job. It currently amounts to about 400Mb.

Some critical programs such as MYOB accounting and Outlook Express email store their data by default in their own program folder or in some obscure place. I have reset those to save their data within the ‘_Documents’ folder tree, so that their data will get backed-up along with other working data.

For each client, I have a ‘Current’ folder tree with folders for each job in progress. Folders for jobs that have been despatched but not yet finalised are moved to the ‘Done’ folder tree. That also helps to prompt me to send invoices! When job data is no longer likely to need further access, I compress the job folder with WinZip to a zip file in the ‘Archives’ folder, and delete the job folder. I also copy the zip file to the archives CD-R for that client, which is kept in packet-writing mode to allow for repeated adding of data. I delete archive zip files from the hard drive after three months. When an archive CD-R gets nearly full, I convert it to CD-ROM format and start another archive CD for that client.

While working, I keep a disk in the 100Mb internal Zip drive and frequently copy files in progress onto that, many times a day. I also save MYOB accounting data to the Zip drive every time I update the accounts. That Zip disk acts as the most immediate and up-to-date backup of work in progress. Since I run many new programs to write about, there is always the risk of a bad one crashing the system, so I need to have very current backup copies. I delete files from the Zip drive when work on them is finalized and they have been backed-up elsewhere.

I keep a CD-RW disk that is a mirror copy of the ‘_Documents’ folder tree. I update that every day before switching off. Windows, rather strangely, does not provide any good means for doing that. If you drag the folder name from C drive to the CD, Explorer asks if you want to overwrite any existing files, and then pops up messages whenever it encounters a read-only file or one of the same name whether older or newer, and asks you to pick whether to overwrite or not. That requires you to carefully read the details of the two files each time. There is no option for overwriting only if newer. So I have made a Dos batch file that does the job very quickly in one hit using the Xcopy 32-bit Dos program included with Windows. The batch file has only one line:

The options specified there are as follows: /S = Copy sub-folders unless empty. /D = Copy files whose date-time is newer than a same-name destination file. /Y = Overwrite existing files without prompting. /R = Overwrite read-only files. These batch files appear on my desktop as DOS icons, in the guise of PIF files.

My laptop PC is usually connected on my mini-network, and I keep another mirror copy of the ‘_Documents’ folder tree on that. This serves as yet another backup as well as allowing me to work on current tasks while travelling. I have a similar XCOPY batch file to the above but to F drive, which updates the copy on my laptop when it is mapped as F drive.

Finally, every month I create a new CD-R disk of the ‘_Documents’ folder tree, and, less frequently, a CD-R of the ‘_Info’ folder. Those are made in the continuous mode as normal CD-ROMs. At the same time, I delete the CD-RW disk data and start it again with a new copy from the PC. That takes care of the fact that the update copy process does not delete files from the CD that have been deleted or renamed on the PC.

This scheme is proving very satisfactory for me. I hope this description may contain some ideas useful to you in working out your own backup strategy.