Backups for Dummies

Filed By: Robert Moir

Backups for Dummies

One of the best ways to protect your data against loss from problems such as drive failure and viruses is to ensure that you "backup" your computer's hard drive on a regular basis. "Backing up" means that you make a copy of the contents of your hard drive and store it in a secure location separate from your working set of data files. 

Some of you are probably thinking, "Why should I need to do this, when I have my antivirus software to protect me" - that's a very good question, and here is what I hope is a good answer. Quite simply, most anti virus software is reactive, that is, it can only protect against a virus or a Trojan horse after your AV company updates their files to detect it. Oh, and you still need to download the file and apply it to your scanner. If you forget, tough luck!

Now despite the way it all sounds very long and drawn out in my description above, the AV companies all move pretty fast and some of them will have a fix for a new piece of malware available for download within a few hours of seeing it for the first time. Now that is fast, but unfortunately, the malware they are trying to stop may be just as fast, if not faster.

For example, consider the "I Love You" worm, which has been the new piece of hot property on the block at the time of writing. The AV companies could turn a fix within a few hours, which is great, but in that time a lot of people may get hit, and in the case of I Love You at least, the malware wasted no time in attacking systems once it had entered them. At least one very large company you all will have heard of managed to lose a lot of data with this worm, and what saved them from a very grave interruption to their work was the fact that they had a good backup system in place that they were able to use to restore people's work files from.

At this point you may be asking "If I backup my files to save me from computer viruses what happens if I backup the virus too? Won't my files be useless?" - The answer to this is "No" because once you know you have a problem, you can take precautions such as repairing files with a virus scanner as you restore them, or perhaps just not restoring the files you know are infected, but restoring everything else. The point is, with a backup safely stored away, you are in control. You can take your time, ask an anti virus company to analyse suspect files, and wait for them to get back to you with an answer. As long as you keep that backup safe you'll be able to get your data restored safely one way or another.

Another good reason to backup often is because computer hard drives have a finite lifetime, after which they will break down in one of several ways. At this point you need to remove the failing hard drive and replace it with one that works. Unfortunately, all to often the way these problems announce themselves is that one day you turn on your computer and instead of starting as normal it makes a horrid grinding noise and announces it can't find any files. 

Hard drive failure is actually a far more common reason for needing to resort to backups, despite all I've wrote above about viruses. The second most common reason is user error - someone deleting something by mistake. Now I realise these are not very sexy reasons to backup, after all defending yourself against nasty virus spreading hackers sounds much more fun, but either way, you need to backup anything you would have trouble replacing.

Now I know that some people look at the figures for hard drive failures and see how many hundreds of thousands of hours a drive can go without failing and think "well I'll be alright… It'll take forever to use MY disk THAT much" but I want you to remember two things. Firstly is that disk drives are mechanical parts, one of the few mechanical parts inside a computer these days, and like all mechanical parts, they will wear out and fail eventually. It isn't a case of "If" your hard drive fails, it's a case of "When" it fails. The other thing you should remember is that those figures are an average. To get that figure, some drives had to last well past the length of time you see written down. That's good, but as it's an average, some drives had to fail quite some time before the figure you see written down. Don't find out the hard way that your drive is a below average one.

Simple or "Working" Backups 

You can backup your data by making another copy in windows explorer and storing the copy in either another place on the same hard drive, or on a different hard drive on the same computer, and this is common practice before making major changes to something that you might want to "undo" with the minimum of fuss and bother. For example, if I decided I wanted to try out a new layout for this website, or perhaps a new background or a new colour scheme I would copy all of the website files to another location while I made my changes to the main working copy of the website. If I decided that I didn't like the changes, or if I made a big mistake and couldn't undo it easily I could delete the working set of files and replace them with my backup. 

This kind of short-term backup is very useful when you are trying out lots of changes to something like a website or a brochure design, but it is of limited value otherwise. We need to consider other options.

"Data protection" backups 

As I already implied above the sorts of backups where you take a copy of a file and store it on the same hard drive or a different hard drive in the same computer are not nearly good enough to protect our data. You need to take a backup of your data onto removable media such as a zip disk, a tape drive, or perhaps a read/write CD. 

Once you have done that, you should store these backups and never work from them, just use them to restore files if there is a problem. The other thing that some people do which is probably over-kill for the home user but is considered standard procedure for a business is to store these backups somewhere away from the computer they were taken from, possibly in a fire proof safe or even at another site. 

As I say, going that far is overkill for those of us who just like to use our home computer for a bit of letter writing and doing the home accounts and perhaps the taxes, but it would be a good idea to borrow a little from business and store your backups in a drawer away from your computer desk. That way if you spill coffee all over your computer and ruin the hard drive or other components, you know your backups won't be ruined too.

The next thing to consider is what type of backups you want to take. There are two sorts we'll look at, which are full backups and partial backups. Both have their good and bad points as we'll see below, both can help you protect your data, and which one you should be doing is up to you.

Full Backups 

A Full backup is pretty much what it sounds like; it means to take a full, complete backup of the contents of a computer's hard disk. This includes everything, your data, the programs you use (for example your word processor, spreadsheet, games, accounts software), all your data files, and your operating system itself.

The advantage of this is that you know that in the event of a problem on the computer you can restore everything to exactly how it was before things went wrong. Your data will be safe, of course, but also your programs will work as they always did without you needing to invest time setting them up and changing settings again. The downside to this is that you need to have lots of space available on whatever it is you use for backups as the operating systems and programs tend to take up a lot of room, typically far more room than your data files.

Because of this, full backups are usually the thing to do only if you have a very spacious backup system, such as read/write CDs, a large capacity tape drive or a DVD Ram drive. It may also take some time to backup all of a large hard drive. That can't be helped, and it's a price you'll just have to pay if you want piece of mind. 

Partial backups 

A partial backup is again pretty much what it sounds like. You backup just your data and perhaps one or two other vital things on your computer such as program or operating system settings that you would find difficult or especially time consuming to have to do again. For example you would backup your data and perhaps any patches you downloaded from the Internet for your operating system and programs.

The advantage of this method is that it is fast and flexible. You can backup your data whenever it suits you, and things like patches you can do whenever you make changes to your system, so it can be a much faster method than the full backup system which requires you to sit and wait while your computer backs up everything in site. It also means you can backup onto much smaller systems than you need for a full backup. For example, I can do my backing up onto just a few 200Mb zip disks. People who are starting out might even get away with backing up their data onto a few floppy disks, at least for a while. 

The downside to this kind of backup system is that when you have to restore a whole system it takes more work on your behalf than you would have to do with a full backup, as you have to firstly have to set up your operating system and programs from scratch and then load up your backups and apply all the patches and finally, copy over your data. 

Quite a few people use a combination of the full and partial backup systems by making full backups of their system whenever they make a major change to the system set-up, and at other times just perform a partial backup of their data files. This can be a little confusing if you lose track of what you are doing, but it's the only way to get the best of both worlds. 

Assessing your backup needs

(Assumes reader is administering a server and talking about network backup needs.)

Simply put, a full backup every day is the best sort of backup you can have, unless you have specialist needs, and a specialist budget. If you need specialist backup strategies, and have that specialist budget you are better off talking to a specialist than you are reading on ;-)

How close you can get to this ideal depends on how much money you want to spend on a backup solution, and that depends on how valuable the data is to the company - simply put how much risk is the company prepared to take with its data? And how much money for reducing those risks does the company feel is appropriate?

That's all it takes to cost out your backup needs. No black magic, no need to hire a consultant, just simple risk management and a bit of arithmetic. Figure out how much per day the cost of having your data unavailable would be, and compare that to the estimated costs of buying backup software and hardware and maintaining it.

To give you an idea, here is the backup strategy for an organisation supporting about 3000 users: 

On our busiest servers, the ones that actually hold our customer's data, and our main server for our staff, we do a daily full backup of every bit of data on every disk to a DLT tape drive using CA's Arcserve. We have a monthly cycle of tapes for these servers (enough tapes to backup the server every day for a month without reusing tapes) and every month we "archive" a tape to allow us to roll back and recover data from a point in time as far back in the past as we need. The monthly range of tapes is stored in a small fire safe in a part of the building away from the servers and the archive tapes are stored in a large safe in a different building far away from the server rooms.

Now this is expensive, both in terms of money and time, but it means that we have a full backup of the servers, and in the event of disaster we can restore data to just about any point we choose.

If you just have one or two servers with a moderate amount of data, and you can't justify the cost of something like Arcserve and DLT tape drives  you can start out with a more modest tape drive and the NT Backup program
that's included with NT Server.  I'd suggest always doing backups AT LEAST once a week, make them all full backups and archiving tapes from your backup cycle every month or so. 

  • Archive a different tape in the cycle every time you do this so they all get renewed before they become unreliable.
  • Check your backup logs daily. Make resolving any problems that are preventing a backup from running a priority.
  • Figure out how you will restore your data in the event of a disaster. Restoring is as important as backing up, which sounds obvious enough but is a point often missed by system managers who make elaborate backup plans, but just shrug when you ask to see their restore plan. If you have a backup you can't restore then what was the point of backing up? Have a disaster recovery plan, document it carefully, and if at all possible rehearse it every now and again. 
  • Make sure there are no hidden surprises like there being only person who has the keys to the safe or the password for the backup software because in the real world you'll need that key or password urgently one day after they go on a 3 month vacation. By the time they come back your business will probably be bankrupt (Don't laugh, I've known this to happen)

Top