You are on page 1of 26

Backups and Disaster Recovery

"What, me worry?" -- Alfred E. Neuman


One of the things that I harp on the most about when it comes to preventive
maintenance and system care, is the importance of regular, reliable data backup. No
matter how well you treat your system, no matter how much care you take, you cannot
guarantee that your data will be safe if it exists in only one place. The risks are much
greater than most people realize.
How important is your data to you? You can respond to this question with words, but
the steps you take to protect your data are the real answer. I find it troubling when
people lose large quantities of data--because they have no backups--and then they get
very upset, talking about "how important that data was!" If it's important, why wasn't
it backed up? If it matters enough to get upset over losing, it's worth protecting, and
backups are an essential part of data protection.
There are many reasons that people neglect doing backups:
• They don't understand how important they are, because they haven't had a
disaster happen to them (yet).
• They don't know how to do them.
• They forget to do them because they don't have a routine for doing backups.
• Doing the backup is a time-consuming chore and they can't be bothered.
This chapter takes a full look at the matter of backups, and discusses how to do them,
how to set up a backup program, and what to do in the event of a disaster. Part of the
focus is on making backups easy to do so that you will remember to back up regularly
without it taking an inordinate amount of time and energy.
Warning: Recent research has uncovered a potential problem with corruption of
short file name aliases when restoring files with long file names using some
backup software. This phenomenon is discussed in the article Xcopy Xposed, which
focuses on the problem as manifested when using the Xcopy command. You should
carefully factor this potential risk into your selection of a backup methodology.

A Mental Exercise To Underscore the


Importance of Backups
Most unfortunately, there are some things that people only take seriously after they
have experienced personally the pain that results from not taking them seriously.
Backups definitely fall into this category. Most people are relatively nonchalant about
them until disaster strikes--thereafter, they are much more diligent about backups (but
after the damage is done.) Despite the difficulty in getting people to learn from others'
mistakes, I'm stubborn, so I will try anyway in this section. :^)
Here's a mental exercise that you can do to help you understand how important
backups are. Take a look at your PC and think about what is on it. Think about your
data and your programs. Consider how much time it took to create the data, and to set
up and tweak your PC so that it works the way you like. Now imagine that one
morning you go to your desk and the PC has vanished without a trace. What will you
do?
Let's suppose you had insurance on the hardware, and a week later a new PC shows
up at your door with a fresh new, clean hard disk. Now what? Most people who ask
themselves this question seriously, begin to take backups much more seriously.
(Fortunately, for most people the exercise is only a mental exercise, but don't think it
can't happen to you in the real world.)
Recovering from a disaster such as a total disk crash or theft of a PC box can be a
very traumatic event, much more than most PC users realize. This is true even if
backups exist; when they don't exist the situation is much, much worse. The pain of
recovering from a disaster is almost always very high, and the cost is primarily in the
time required to recreate the lost data. For even a small business, this can run into the
thousands of dollars very quickly.
If you still aren't convinced, consider this report from the University of Texas Center
for Research on Information Systems. Of the companies that lose their data in a
disaster:
• 90% are out of business within two years...
• Nearly 50% never reopen their doors at all after the disaster!

• The Risks To Your Data


• If you do any sort of reasonable amount of computing, it is only a matter of
time before you some day need access to backups of your data or programs.
There are many different risks to your data; most people just think of the
infamous, dreaded disk crash. This is a real risk due to the technology used in
data storage, but there are many other ways that you can easily lose data on
your PC. In fact, the list of risks below is far from exhaustive, though it covers
the most common problems.
• Some backup methods protect against all of the risks below, while some protect
only against one or a few of them. See this section for a comparison of the risk
protection offered by various backup methods.
Hardware Failure
The risk of hardware failure is the most commonly talked-about reason to perform
backups. Indeed, nothing will jolt someone into realizing the importance of backups
more than an unrecoverable hard disk failure. Since the hard disk stores your main
programs and data, it is the hardware whose failure hurts the most. It is also what gets
the most attention, and rightly so.
However, there are other hardware problems that can cause permanent data loss, and
some of these can be rather hard to figure out, since they don't seem like they should
be responsible for the problem. Here are just a few:
• Memory Errors: With so many systems today running without error detection
or correction on their system memory, there is a chance of a memory error
corrupting the data on the hard disk. It is rare for it to happen, but it does
happen.
• System Timing Problems: Setting the timing for memory or cache access too
aggressively, or using a hard disk interface transfer mode that is too fast for the
system or device, can cause data loss. This is often not something that will
generally be realized until after some amount of damage has been done.
• Resource Conflicts: Conflicts resulting from peripherals that try to use the
same interrupt requests, DMA channels or I/O addresses, can cause data to
become corrupted.
• Power Loss: Losing power at the wrong time, such as when you are doing
sensitive work on your hard disk, can easily result in the loss of many files.
Software Failure
It is possible for data to be lost due to software bugs, or even just poor software
design. For example, a program might have a problem where it crashes upon saving a
file. Many programs, when saving a file over top of an older file with the same name
(such as when you select "Save" to update the current document you are working on)
will first save the new file under a temporary name, and then rename it to the correct
file name when the save is completed. But others may remove the old file first, so that
if the software crashes during the write process, the old file will be lost as well.
Some software bugs may be even more damaging, even causing the loss of files
unrelated to them. This doesn't happen very often, fortunately.
File System Corruption
There are many ways that the file structures used to contain programs and data on the
hard disk can become damaged. In some cases, this corruption can result in data loss,
especially if the disk is not maintained properly and the file system scanned on a
regular basis.
Accidental Deletion
Since computer users are human, they make mistakes. One of the most common is
accidentally deleting files from the hard disk. There are many different protection
mechanisms and "undeletion" utilities that can help recover from this, but sometimes
you delete a file and then remember a few days later that you really need it, and in this
situation a backup is often the only thing that can save you.
There are other ways that you may accidentally delete files, maybe without even
realizing it. Let's suppose that you have twenty files named MEMO01.DOC through
MEMO20.DOC in a directory, and you want to move them to the directory
C:\OLDMEMOS. So you type:
• COPY MEMO*.DOC C:\OLDMEMOS
• DEL MEMO*.DOC
But let's suppose that you made a mistake and the C:\OLDMEMOS directory doesn't
exist, or you spelled it wrong when you created it. If this is the case then the copy
command above will not be able to copy the files to a C:\OLDMEMOS directory.
Instead, it will interpret the "C:\OLDMEMOS" as specifying the name of a file instead
of a directory, and will copy the contents of all twenty files to a regular file named
"OLDMEMOS" in the C:\ root directory. The end result will be a single file
containing the data in each of the 20 files, concatenated end-to-end. (You may be able
to recover from this if these are plain text files, but otherwise you are in trouble.)
When the "DEL MEMO*.DOC" command is executed, it will remove all the files
from the directory they started in. You may not even realize that this is happening. Of
course being careful can avoid this sort of problem (be sure of what you delete!) but a
backup can save you from the occasional accident.
Accidental deletion can occur even within a file. In creating this web site, I started
first with a very large, single document. On more than one occasion, I deleted sections
of the text that I thought I didn't need, and was glad that I had archived older copies of
the file that I could refer to, to restore the parts that I found out I actually needed later
on.
Virus Infection
Viruses can easily cause the loss of data, in many ways. This includes loss caused
directly by the virus and also program damage that results from efforts to remove
viruses from a system.
Theft
Many PC users don't consider the possibility of theft as a danger to their data
(although most people who use notebook PCs certainly do!) Even for desktops, there
is always the risk of the entire PC box disappearing one day if you have a break in, for
example. It is important to keep this possibility in mind, as some kinds of backups do
nothing to protect against theft. To my knowledge, insurance that covers theft of
computers will allow you to replace the system, but cannot pay to recreate lost data.
Sabotage
Generally applicable only to PCs in the work environment, sabotage by disgruntled
employees is a growing problem. A carefully-constructed backup program is the only
protection against a knowledgeable but angry person who is determined to
intentionally cause data loss (and sometimes, even backups aren't enough.)
Natural Disaster
Fire, flood, earthquake, mud slide, hurricane, lightning strike, you name it: all can
result in the destruction of your PC and everything on it. For many people getting
back their data will be the last thing on their minds when this happens, of course. For
others, getting back up and running on a PC may be something they need to do right
away. Also remember that insurance may cover the PC, but will not cover the data
that is on it.

Backup Methods, Devices and Media


There are many different methods that you can use to back up the data on your hard
disk. The primary difference between these methods is the device and medium that is
used to store the backup. Different media have different characteristics, such as
capacity, speed, ease-of-use, universality, etc. This section takes a look at some of the
different choices you have for backing up your hard disk.
Media Size Matching
One very, very important factor to consider when looking at backup alternatives, is
matching the size of the backup medium to the amount of data you need to backup. As
hard disks continue to increase greatly in size, it becomes more difficult to find
backup solutions that can handle the entire contents of a PC using a reasonable
amount of media. It is essential that the size of the backup medium be matched to the
size of the data being backed up.
It is tempting to ignore this issue as unimportant, but my experience (and that of
others) is very clear: the more disks or tapes it takes to perform a backup, the less
likely it is that they will be done on a regular basis. The reason is simple: when it
takes a lot of media to back up the disk, backup becomes a chore, and when it
becomes a chore, people avoid doing it.
In fact, the best situation of all, and one that I recommend, is a backup solution where
the entire contents of the hard disk can be stored on a single backup tape or disk. What
this buys you is the ability to do unattended backup. You start the backup, and then
leave to do something else. When you return, the backup is done. If the contents won't
fit on a single cartridge, you have to intervene at some point to change media. This
changes backup from something simple to something complicated. Don't
underestimate capacity when comparing backup solutions.
Since you don't always need to back up the whole hard disk at once, a backup unit that
can't hold the entire hard disk, but can hold most of it, will normally suffice. If you
can store half the hard disk on a single cartridge, you can do unattended backups of
half the drive at a time. Less ideal than doing the whole system at once, but still quite
tolerable, and you maintain the ability to do unattended backup.
Let's take an example look at the Iomega Zip drive. This is a removable storage device
with a capacity of about 100 MB. A very useful drive, it stores data reliably, and if the
parallel version is used, can be extremely handy for transferring information between
PCs. However, it is also advertised as being great for backup. Well, I guess it depends
on your system, but mine has over 3 GB of programs and data on it. Many people
have over 1 GB on their hard disks. For this amount of data, the Zip drive is not
suitable as a backup device. Backing up 3 GB of data onto Zip disks would take 30
disks (costing several hundred dollars, mind you) and the disk swapping would be
annoying enough to guarantee that the backup was rarely if ever done. A 100 MB
drive can be useful for archiving parts of a large disk, but it isn't a solution for a
general backup, because you can't even do half or even a quarter of a modern hard
disk on such a unit.
Removable Storage Drives
In the last few years, a whole new class of storage devices has sprung up and become
very popular--removable storage drives. While these have existed in various forms for
many years, it is only recently that they have taken the market by storm. Their
popularity rests in their ability to provide removable storage at a reasonable price and
with good performance. Many of these drives are also very suitable to use for
backups.
There are so many different drives, and they differ in so many different ways that a
complete look at all of them (at least one that would be fair) is far beyond the scope of
this section. (I hope to add a full chapter in the Reference Guide that will look at these
drives at some point in the future.) Looking at the various characteristics of these
drives, we see that many of them are suitable for backup purposes, but some really are
not. The drives fall into several categories, which I will look at briefly:
• Large Floppy Disk Equivalent Drives: This would include the Iomega Zip
drive, Syquest's EZ-135, the LS-120 120 MB floppy drive, and a few others.
These devices are suitable for backup only if you have a small hard disk, or
have the diligence and patience to do attended backups or large numbers of
partial backups. As hard disks increase in size to 4 GB and beyond, trying to do
backups to a device that is only a little more than 100 MB becomes impractical,
and quite expensive. The reliability of these devices is quite good, although
they are proprietary and not very universal. Their performance is general poor
to average.
• Removable Hard Disk Equivalent Drives: This category includes devices
such as Iomega's Jaz drive, Syquest's SyJet, and various kinds of phase-change
and magneto-optical drives. These are much more suitable for use as backup
devices due to their larger capacity, but even here things are becoming
stretched, since even 500 MB to 1 GB is becoming inadequate for unattended
backups. These drives have generally much higher performance than the
smaller drives, and much higher price tags to go with them. Reliability is
usually good, and the drives are still proprietary.
• CD-Recordable: These are write-once read-many drives with a capacity of
about 650 MB, described in detail here. Despite the fact that the disks are not
reusable, some people actually use them for backup, now that the price of blank
disks has gone down to around $3 a piece or so. This is a very expensive way to
do backups; it does give you the advantage of being able to refer back to
historical snapshots of your disk for a long time, but you're really going to pay
for it. The cost will discourage most people from doing backups often enough.
The capacity is on the small side at 650 MB. One great advantage is that the
backups are readable by any CD-ROM drive. I do not recommend CD-R for
routine backups, because the cost of media over time is excessive. This will
discourage you from doing backups on a regular basis, which is the last thing
you want.
• CD-Rewriteable: This drive is really in the same category as the removable
hard disk equivalents listed above. CD-RW has a lot going for it as a general-
purpose medium, because of its flexibility: its media are reusable and it can
also burn CD-Rs that play in most CD-ROMs or audio CDs as well. But as a
strictly backup medium, I don't think it has a lot to recommend it. It is not
inexpensive, the capacity is only so-so at 650 MB, and the CD-RW disks are
essentially proprietary since only newer CD-ROM drives will read them. It's
certainly usable for backup, but other options may be better if you don't need
the other advantages of CD-RW.
Removable Hard Disks
An interesting backup method that most people don't know about is the use of
removable hard disk drives. Here's how it works. You purchase a special kit that
includes a mounting kit that you install into an external drive bay in your case. You
also get special adapters (sometimes called carriers) that attach to regular, internal
IDE hard disks. This allows you to insert and remove these internal disks through an
external drive bay, turning your regular IDE hard disk into a sort-of removable drive.
It works in a way similar to how a removable car stereo works.
To allow for backup flexibility, you will want to have more than one drive. One way
to handle this is to buy several identical hard disks and put them all in carriers, so you
can swap them easily. Or, if your BIOS supports hard disk autodetection, you can use
different types of disks and the system will reconfigure to use whatever is currently
placed in the drive bay each time you reboot.
This type of system, even though it sounds strange, can actually be a quite viable
backup solution. While it seems that it would be overly expensive to buy hard
disks just for backup, the price per gigabyte of hard disks is actually very
comparable to drives such as the Iomega Jaz--in many cases, much less. This is
especially true if you buy slightly older, smaller disks. A removable hard disk also
has the following other advantages: very high performance, random-access
capability, standard interfaces and exchangeability, and excellent reliability.
This type of scheme has some disadvantages as well of course. Compared to
something like tape, you lose the ability to buy additional backup media cheaply; to
add 2 GB of more backup storage on a 2 GB tape drive costs about $25; with this
scheme it would be several times more. Hard disks are also fragile; if you drop them
they can be damaged. (Note that both of these disadvantages also apply to several of
the larger removable-storage drives.) The final and perhaps biggest drawback of this
type of scheme is that the disks can only be removed when the power is off; you can
switch media in a standard removable drive on the fly with the power on, but not here.
In-Place Hard Disk Duplication
One backup solution that some people who have more than one hard disk use is to set
up their system so that they use one hard disk for their programs and data, and the
other one as a backup. They then copy the contents of the one hard disk to the other
on a regular basis, using a file- or disk-copy tool.
This scheme is interesting and can be useful in some ways, although it has
some severe limitations as well. Its advantages are that it is simple, and that it is
automatable; you can set it up to do this on a nightly basis and know that everything
on your hard disk will be duplicated automatically without you having to remember to
do anything. This increases the chances that the backup will be done. The
performance of the disk-to-disk copy will be very high, and the cost is pretty much
reasonable.
The drawbacks of this sort of a scheme however are significant. First, it doesn't
protect against very many of the risks to your data--it is not going to help much
against theft, fire, sabotage, many types of viruses, and even some types of hardware
failure. Second, you can only have a single backup, which makes the whole system
very vulnerable--if you make a copy of the whole disk every night, what happens if
you only notice a problem three days after it wipes out some of your data? Finally, the
temptation is large to use the second drive for more data and discontinue the backup
procedure when the first disk gets filled up,.
Overall, I don't recommend in-place hard drive duplication as a standalone backup
procedure. It can be useful when supplemented by a removable backup system.
Network Backup
For PCs on a network, backup over the network is a viable alternative to using
removable drives. This type of scheme is sometimes used in small- to medium-sized
as a way of protecting PCs without the expensive of tape drives or removable storage.
The idea is fairly simple: copy data from one PC to another over the network.
Duplicating each PC's information provides a way to protect each individual PC.
In a way, this type of backup is most similar to in-place hard disk duplication in terms
of how it works. It is simple in the same way, and can be automated. It addresses
some of the concerns about that method: there isn't the same single point of failure in
terms of virus attack or hardware failure. However, depending on the location of the
two PCs, theft, disaster and sabotage can still be a big problem: if the two PCs are
sitting on different desks in the same office, you haven't gained much to protect
against these threats. Also remember that file-infector viruses can travel over a
network.
An even better use of the network, becoming popular in many corporate
environments, is to use a centralized removable storage backup device in conjunction
with the network to back up all the PCs automatically. Consider a local-area network
with 10 PCs, each containing 1 GB of programs and data, connected to a server with 5
GB of storage and an 8 mm tape drive with a capacity of 10 GB. Using compression,
this drive can hold the entire contents of the network on a single tape, and using
network backup software and the right operating system, the network can be used to
back up all the PCs every night, automatically. This is an excellent backup system that
allows the desktop PCs to reap the benefits of the tape backup unit without everyone
having to remember to do backups.
File Archiving
A supplemental backup method that I use, and recommend that others consider, is
simple file archiving. What I mean by this is simply making backup copies of files
that I use periodically, in case I need them later on. When I am working on a large
document over a period of weeks or months, this document is changing far more often
than the other files on my hard disk. I want to therefore back it up more often, in case
I make a mistake in the document or my program decides to munch it. It's not
practical to break out my tape drive several times a day, so I do this by simply
copying the file to another location on the hard disk once in a while.
This is a limited form of in-place hard disk duplication, which is not a complete
backup solution. In fact, this system only really protects well against accidental
deletion, which is why I call it supplemental. However, it can still be very useful to do
this. A simple way (which is what I have done) is to create a directory on your C:
drive (or wherever) called BACKUP, and then create a simple batch file containing
this line: "COPY %1 C:\BACKUP". (Batch files are files that contain commands that
are executed when you run them.)
Save the file as say, BU.BAT, and store it in a directory that is in your file execution
path, such as C:\WINDOWS\COMMAND. Then, to back up FILE.TXT you just type
"BU FILE.TXT". Make sure to clean the directory out periodically.
Comparison of Backup Method Data Risk Coverage
Some of the backup methods and devices described in this section do a much better
job than others of protecting against the risks to your data. The table below shows a
general summary of how the various methods stack up in terms of protecting you from
the hazards that threaten your valuable data. Remember that this is just a general
guideline; also remember that some of these risks are much more common than others
are, depending on how you use your PC:
Data Risk Floppy Tape Removable Removable In-Place Network File
Disks Drives Storage Hard Disks Hard Disk Backup Archiving
Drives Duplication

Hardware Moderate to Moderate to


High High Moderate High Low
Failure High High

Software
High High High High Moderate High Low
Failure

File System Low to Moderate


High High High High Moderate
Corruption Moderate to High

Accidental
High High High High High High High
Deletion

Virus Moderate to Moderate Low to


Moderate High Moderate Low
Infection High to High Moderate

Moderate to Low to
Theft High High High None None
High Moderate

Sabotage High High High High Very Low Low None

Natural Low to
High High High High None None
Disaster Moderate

Comparison of Backup Method Characteristics


In order to help you understand how the various backup methods compare to each
other and help you make a decision about which one makes the most sense for you, I
have enclosed the chart below. It evaluates the various backup methods in several
different important general areas. These areas are:
• Capacity: How much does the medium store, relative to current average hard
disks? Is it likely that an average user's hard disk will fit onto one or maybe two
pieces of media for the device/method?
• Automatability: How possible is it to fully automate the backup so that it
happens without any user intervention at all?
• Minimized Startup Cost: How well does the method minimize startup costs
for hardware or software?
• Minimized Media Cost: Does the method allow additional backups at a
reasonable cost?
• Expandability: How possible is it to do more backups, or add additional
backup media if needed?
• Reliability: In general terms--because this varies widely based on the type of
device for items such as tape drives--how likely is it that if you have a disaster
and need to restore from the backup, it will work for you?
• Simplicity / Convenience: How easy is the method to use? Is there any
difficulty associated with the method that would tend to discourage doing
backups?
• Universality: How common is the hardware used for the method? If you
needed to use the device in five years, how likely is it that you could find
support or additional media for it?
• Performance: How fast is the hardware and software used for the method?
How much time will it take to do a backup?
• Routine Potential: In general, how likely is it that, using this method, someone
is likely to settle into a backup routine and stick with it?
Here is the chart; for all items, "High" is better and "Low" is worse (I inverted the cost
items so that high had a better meaning and low worse, to keep things consistent,
sorry if it is confusing):
Removable In-Place
Floppy Tape Removable Network File
Characteristic Storage Hard Disk
Disks Drives Hard Disks Backup Archiving
Drives Duplication

Very Low to
Capacity High High High High --
Low High

Automatability Low Low Low Low High High High

Minimized Very Low to Low to


Moderate Moderate High Very High
Startup Cost High High Moderate

Minimized Moderate Low to Very


Low Moderate Very Low Very High
Media Cost to High Moderate High

Moderate to Very
Expandability High High High Low High
High High

Low to Moderate to Moderate to Very


Reliability Low High Very High
High High High High

Simplicity / Moderate Moderate to


Moderate Low Very High Moderate High
Convenience to High High

Very Low to Low to


Universality Moderate High High High
High High Moderate

Performance Very Low to Low to Very High Very High Moderate Very High
Low Moderate High to High

Routine Very Moderate to


Moderate Moderate Moderate High Very High
Potential Low High

Note: For a more complete picture on how the various methods stack up, make
sure to check out how they cover the various risks to your data as well. Many of
the methods that look very attractive based on what you see above actually don't
protect against enough of the data risks that you use backups for, to make them truly
viable options.

What To Back Up
To ensure that your backups are performed properly, in a way that ensures that you
are protected without taking so much of your time that they become a chore, you must
determine what files to back up and how often to back them up. Some files will need
to be backed up more often than others. This section takes a look at what files you will
want to include in your backup routine, and also the ones you will usually want to
exclude.
Full, Selective and Incremental Backups
There are several different ways that we can select files for backup. The way that you
will want to use depends on how you use your system, how often your files change,
and your chosen backup method. Many people will in fact end up using a combination
of these different techniques:
• Full Backup: Nothing complicated here, a full backup is done
by selecting all the files on the hard disk for backup.
Onlyspecial files that should not be backed up at all are left
out. This is the simplest type of backup, and yields the most
complete backup image, but it takes the most time and media
space to do.
• Selective Backup: In a selective (or partial) backup, you
select specific files and directories to back up. This type of
backup gives you more control over what is backed up, at the
expense of leaving part of the hard disk unprotected (unless of
course, you simply do another selective backup later on that
covers the area you skipped the first time). Selective backups
make sense when some files are changing much more rapidly
than others, or when backup space is limited, although in many
cases doing an incremental backup is better and easier.
• Incremental Backup: If you perform frequent backups, as
you should, you may find yourself backing up the same files
over and over, even ones that do not change over time.
Instead, you may want to consider a mix of full backups and
incremental backups. An incremental backup is one where only
the files that have changed since the last backup are selected.
It is like a selective backup, but the files are selected based on
whether they have changed recently or not, instead of an
arbitrary selection based on directory or file names. This gives
the time- and space-saving advantages of a selective backup
while also ensuring that all changed files are covered.
Incremental backups are supported by most decent backup software. They work using
the archive bit that exists for each file and directory. The backup software looks at this
bit to determine what files have been changed since the last backup, selects them for
backup, and then clears the bit for all the files it backs up. If any files are changed, the
software sets the bit again so on the next incremental they are again selected, and so
on. You must rely on this bit being managed properly, and I don't always like to do
this.
Warning: Programs that change files are not required to set the archive bit. Most
well-behaved software will do this, but you cannot absolutely guarantee that all
files changed since the last backup will be caught by an incremental selection.

Which type of backup you do depends, again, on what is important to you, in terms of
time, media cost, and also ease of restoration.Restoring a system that uses incremental
backups can require more steps, as first the full backup has to be restored and then the
incrementals, one after the other. You also don't have the redundancy (just in case you
need it) that you have when your backup scheme uses only full backups. If you can fit
a full backup onto a single media set and it doesn't take too long to do, I still prefer
this over incrementals, for this reason. It's just more secure to know that everything is
on one tape. The scheduling of backups is discussed here.
Backing Up Programs and Data
Most files on a typical PC can be broken down, loosely, into being either programs or
data. These differ of course in many ways, but not least is in how they should be
considered for backup.
In general, data files should always be backed up. These are the fruit of your labor,
and cannot be replaced! Every backup should include all of your data, either via a full
backup of the entire hard disk, a selective backup that includes the directories where
the data is, or an incremental backup. (Having all your data in a handful of directories
instead of scattered all over the hard disk is useful, for this very purpose).
Programs are a somewhat different story for two main reasons: first, they are static,
meaning once installed they do not change (with a few exceptions). Second, they are
recreatable; if your Microsoft Office directory gets wiped out, you can reinstall it from
your original CD-ROM disk. The combination of these characteristics suggests that
backing up programs is less important than backing up data, and this is true. Programs
do not need to be backed up as often as data does.
However, you should also remember that setting up and configuring a PC today takes
a considerable amount of time. Tweaking all the operating system settings, installing
all the software, and modifying parameters to get everything working the way you
want can take the better part of a weekend. While you can reinstall Windows 95 and
all of your applications if you have to, this is not something that is going to be fun to
do. You will also have the problem of remembering what you changed from the
standard defaults. For these reasons, installed programs should definitely not be
ignored when looking at backup.
Note: Some PCs ship with their software preinstalled on the hard disk and no
original disks or CD-ROMs! This is a poor practice and I recommend that people
avoid buying from companies that do this, since it makes it very hard for you to
reinstall software if you need it in case of disaster. If you have no original disks, your
installed programs should be treated as just as unrecreatable as your data. Incidentally,
many PCs that come only with preinstalled software have utilities that will let you
create install disks from what is stored on the hard disk, sort of a "reverse install";
consult your owner's manual.

I believe that a full backup of the entire hard disk is still the best type of backup, for
the simple reason that it is the safest and easiest way to get you back where you were
before a catastrophic data loss--and that is the exact purpose of backup. If you don't
have the backup capacity to do the whole hard disk, then consider a combination of
full and incremental backups, which will preserve any changes you make to your
programs while saving media.
Files Not to Back Up
There are some files that should never be backed up. The main reason is that some
files do not contain anything that would matter if they were lost. Another reason is
that some files are representations of programs or data that are better backed up in a
different form. While virtually all regular files should be backed up, in most cases the
following kinds of files should be excluded from routine backups:
• Swap Files: A swap file is a large file that is used by the
operating system for virtual memory. When the system needs
more memory than actually exists in the PC, it creates a virtual
memory space and applications "share" the real memory by
swapping pieces of memory to the hard disk. The swap file
holds these pieces; the process is described in much more
detail here. This file can be quite large, sometimes exceeding
50 MB in size depending on the system. Since it does not
contain any real data, but rather is a placeholder for
information in memory while the PC is running, there is no
point in backing it up.
• Compressed Volume Files: If you use volume-based disk
compression, the compressed volume that you see mounted as
a drive letter is stored on the host disk as a single file called
a compressed volume file or CVF. All the files on the
compressed disk are in this file. If you use compressed
volumes, the files on them should be backed up individually
from the compressed volume. The big CVF file on the host disk
should not be backed up, since it contains the same
information but in a way that is much more difficult to restore.
Most newer software will in fact automatically deselect the items above, unless you
override and tell it you want them included anyway. Many types of backup software
will also let you select classes of files, by file type, that you want to exclude for
whatever reason.

How To Back Up
While backing up your data is in some ways a simple matter--"just do it!"--there are in
fact some special techniques that can come into play to make backups more effective
and less of a hassle. This section takes a look at specific techniques and considerations
for performing backups, some of which you might not think of. This includes a
discussion of backup timing, scheduling, media storage, and how to ensure that your
backups work, and will protect you in the event that you need them.
Backup Timing
Selecting a time of day to perform backups is a matter of personal choice. It depends,
as usual, on how you use your PC, and also on how long it takes for you to perform
backups. Most people prefer to do unattended backups, and therefore, set their
backups up to run when they are not around. The most common times to run backups
therefore are:
• Overnight: My personal choice, and the preference of many
other people, is to start a backup before going to sleep, have it
run overnight, and then see the results in the morning. This
makes sense because the PC isn't being used, and a full
backup of a loaded system, including time to
perform verification, can take several hours.
• During the Day: If you work during the day you can set up
your home machine to perform backups while you are at the
office. This is really not much different in concept from the
overnight backup; both use slack time where the PC isn't being
actively used, to perform backups. I don't do this myself
because I run routine maintenance on my PC during the day.
Warning: I only recommend backup overnight or while away from home for
backup systems that are established and known to work well. Do not attempt
your first backup on a new drive overnight, because if there is some problem, you
won't find out about it until 8+ hours later. A misconfigured or defective tape drive
can sometimes "shoeshine", meaning it constantly runs the tape back and forth over
the surface of the tape head. This can be very damaging if allowed to continue for
hours at a time!

Most backup software can be set to run at a specific time and day of week as well, if
this is convenient for you. Remember to watch out for interference from other
software.
Backup Software
An important part of the backup puzzle is using the right software. The difference
between good and mediocre backup software can be the difference between backups
that are reliable and easy to use and ones that are not. The difference between
mediocre software and bad software can be the difference between backups that
restore properly when you need them to, and those that leave you high and dry!
This is a hardware site and I am getting pretty far afield with this long look at backup
as it is, so I am not going to try to get into too many specifics on backup software
(which I could write quite a lot about). You have to carefully compare the different
packages out there and get one that meets your needs, much as with any other piece of
software. The various features of backup software can be confusing however, so I will
list below the types of capabilities you will want to look for in backup software, to
help you make a good choice when you look for a package.
Beware that here, as in many other places, you sometimes only get what you pay for.
Many backup devices ship with basic backup software, provided as a courtesy by the
hardware manufacturer. In many cases these are functional but stripped-down
versions of commercial packages. They will usually work, but may not be nearly as
full-featured as a package you would buy at the store. The best thing to do is to try
any software that comes with your device; if it meets your needs then you don't need
to buy anything else.
The following are abilities or features that you may want to consider carefully when
looking at PC backup software (not listed in any particular order):
• Wide Device Support: Backup software varies significantly in
its ability to support backup devices. Generally speaking, it is
more difficult to find software support for newer devices than
well-established ones. Some software companies will make
software updates available for their users to provide expanded
support as new drives hit the market; others will not. Do
remember that while support for more devices gives you more
flexibility, ultimately the only device you really need support
for is the one that you are actually using.
• Operating System Support: The software should support all
of the features and requirements of the operating system
under which it runs. This means, for example, that Windows 95
software should have full support for long filenames, backup of
the Windows 95 Registry, and backup of FAT32 partitions.
• Backup Type Selection: All good backup software will let you
choose between doing full, selective and incremental backups.
Better ones will let you select files and directories based on
search strings or patterns.
• Media Spanning: The software should provide proper support
for backing up to multiple pieces of media in a media set. So if
you did a backup to Zip disks and the data took up 250 MB, the
system should prompt you when it is time to switch disks, etc.
Strangely, some poor backup software has problems with this.
• Disaster Recovery: A very important feature, and one that is
often found only on more expensive products (as opposed to
the freebies that come with many tape drives) is support
for automatic disaster recovery. With this type of software,
sometimes called one-step recovery or single-step restore or
similar, a floppy disk is created with a special recovery
program that will let you restore your system simply. Without
this feature, you often have to reinstall the entire operating
system before you do a restore, which can cost a lot of time
and cause a lot of problems.
• Scheduling and Automatic Operation: Depending on how
and when you do your backups, it can be very helpful to have
the software run automatically at a preset time. Most of
today's software will support this.
• Backup Verification: Every decent backup package will allow
you to enable a verification mode. When active, the software
will read back from the tape every file that it backs up and
compare it to the file on the hard disk, to ensure that the
backup is correct. This is important to ensure that your
backups are viable.
• Compression: Good backup software will give you the option
of enabling software compression, possibly at various levels, to
enable you to save space on your backup media.
• Media Append and Overwrite: You should be able to set the
software so that you can control easily what happens when the
software starts a backup of a tape that already contains a
backup set. You should be able to tell the software to always
append to the tape, always overwrite it, or prompt you each
time to let you select.
• Tape Tools: If you are using a tape backup unit, the backup
software will allow you to do things like formatting, rewinding,
retensioning or viewing the catalog on your tape. The tape
drive may come with software that does this for your particular
model; it's much easier if the backup software supports these
tools also, however.
• Security: Better software packages will let you password-
protect a backup set so that the password is required to view
or restore from the backup image. (Be very careful before
using something like this, you don't want to lose that
password!)
• Backup Configuration Profiles: You may want to do
different types of backups at different times. For example, you
might have a bunch of compressed ZIP files on one drive and
want them to be backed up with tape compression off (since it
won't do anything anyway) while your regular files on another
drive are backed up with compression. Good software will let
you store different profiles for different types of backups to
save you from having to change things every time.
• General Quality Issues: You should find out about the
general nature of the software. Does it work well? Is it buggy?
Are people having problems with it? What is the warranty?
What is the upgrade policy of the manufacturer? USEnet can
often be of assistance here.
Software Conflicts
When performing backups in a multitasking operating system, it is necessary to be
wary of possible conflicts between the backup software and any other software that
may be running simultaneously. Any programs that are running in the background that
might write to files or directories on the disk can confuse the backup software,
especially when it goes to verify the files it has backed up, because it may find
different files in some directories at the end of the backup compared to what was there
at the beginning.
Another problem is with files that are locked due to another program having exclusive
access to the file. To prevent more than one application from changing a document at
the same time, many applications will lock them out so no other application can use
them. This can cause the backup software to be unable to back up these files.
If you are doing backups from a single-tasking operating system like DOS, you don't
generally have this concern (because you don't have multiple programs running
simultaneously). However, watch out for TSR (terminate and stay-resident) programs
that might be in memory at the same time as your backup software. If you schedule
backups to run at a pre-set time while you are away, you may have a more difficult
time of ensuring that nothing else is running at the time the backup starts.
The solution to avoiding these problems is, in most cases, to simply turn off other
software when you are doing a backup. This is what I do. I close all applications
before I begin a backup, and make sure there is nothing running in the background (I
use Windows 95). I also disable my screen saver, to ensure that it doesn't cause
interference (this can be done by going to the Control Panel, selecting "Display" and
then "Screen Saver"). Another advantage to doing this is that I ensure that the backup
software gets as much of the processor's attention as it needs.
Data Verification
After you have backed up your data to tape, you can feel confident that you have a
copy of your data that you can use in the event that anything happens to your PC. Or
can you? If you're paranoid like me you might want an additional level of security to
help you feel safe. This is easy to do, by using backup software that includes the
ability to perform data verification, and enabling the feature.
The idea behind verification is simple: after the files are backed up, the backup
software reads back the information from the backup media and compares it back to
the original files. This ensures that the backup just made is readable, and that the files
match what was just copied. I recommend that data verification be used, as it is a
simple way of feeling more sure that your backups are working correctly. The only
disadvantage is that it lengthens the amount of time that it takes to perform the
backup, but if you are backing up overnight or while away from the PC, this will have
no effect on you anyway.
There are two different levels of verification that you will sometimes find, depending
on the software you are using. The most secure level of verification is full verification,
where each and every file that is backed up is also verified by reading back from the
backup medium. A lesser type of verification is sampling verification. Here, instead of
verifying everything that was backed up, a sample of what was backed up is read back
and verified. This makes the verification take much less time, but of course doesn't do
nearly as good a job. Usually full verification is easier to find on most software than
sampling.
The ultimate in backup verification is doing a test restore, but this takes both time and
some spare hardware. If you have an extra hard disk or PC, you can take your backup
set and try to restore the backup to this machine. If it works properly, you can feel
confident that your backup will probably protect you when you need it.
Backup Compression
Most backup systems support some type of compression. The idea behind
compression is simple: to save space and allow the backup of more data onto a given
media set. Most software supports compression, and in fact many backup devices
even quote their capacity on the assumption that compression will be used during
backup.
Normally there is nothing wrong with using compression, and in fact I use it myself in
many cases. You do need to bear in mind a few things, however:
• Compressability: Not all files will compress equally well, as
explained in this section on hard disk compression. If you are
backing up a large number of files that are already in a
compressed format, it may make sense to turn off
compression, since it isn't going to do much for you anyway.
• Proprietary Formats: Each software program (or in many
cases, family of software programs made by the same
company) will use its own compression algorithm. This means
that the tape written by one program may not be readable by a
different software package. This is not normally a problem
since most people only use one package on one PC, but it is
something to bear in mind. The backup formats themselves are
reasonably universal if compression is not used.
• Processing Power Requirements: In order to write some
types of backup devices, especially tape drives, it is necessary
for the software to provide to the device a steady stream of
data. This is because tape can only be written to when it is
streaming at constant speed. Compression algorithms take
time to run and can sometimes interfere with this steady flow
of data, causing problems like shoe-shining (where the tape
has to repeatedly back up and restart sections of the backup).
Compression should be turned off in most cases if doing the
backup on a slower PC or if it is suspected that compression
overhead is causing problems.
• Exaggerated Compression Ratios: Watch out for overly-
optimistic estimated compression ratios. Many tape
manufacturers like to claim that their 1.6 GB tapes for example
will hold 3.2 GB "with compression". In practice, I rarely if ever
see any full disk volumes (as opposed to collections of certain
types of files) that will compress at a 2:1 ratio.
Media Storage

Your data backups are exactly as safe as the physical media that contains
them. If you do a tape backup and then leave the tape lying on top of the PC
box, then you are partially defeating the purpose of doing backups. You will
give back the protection that tape offers against risks like theft, disaster or
sabotage.

Backup media should be stored in a safe place, away from the PC. In fact,
the word safe is appropriate, since a locked safe is the best storage
container for backup media. The use of a safe allows the media to be
secured from prying eyes and fingers while remaining in the same general
vicinity as the PC. A fireproof safe is an even better idea, for obvious
reasons. Just make sure you don't lose the combination. :^)

Depending on the type of backup media you are using, you want to make
sure that the storage environment is appropriate. For magnetic media such
as tapes and disks, you want to ensure that the storage area offers
protection from the hazards that threaten them, including temperature,
moisture, dirt, magnetic fields and the like. A plastic box on a sunny window
sill is definitely not a good idea.

Finally, pay attention to the matter of off-site storage. It is a good idea to


ensure that one or more of the backup media sets in the media rotation
system you are using is always be stored off-site. This is important to allow
for safeguarding against total disaster (tornado, hurricane, whatever).

Boot Disks
A very important, but separate, part of your backup strategy should be the creation
and maintenance of boot disks. These disks are used in the event of an emergency
with your system (they are in fact sometimes called emergency boot disks). They are
designed to enable you to quickly and easily correct large-scale software and
operating system problems with your PC, and to allow you to reconfigure or update
your system without being totally dependent on the contents of your hard disk.
This section discusses boot disks and how they work, and provides details on how to
make and use them effectively.
Importance of Boot Disks
You rely on your hard disk to boot your PC each day, but what would happen if
something happened to your hard disk to make it unbootable? There are many
different situations that can cause this to happen, ranging from hardware failure to
erroneous operating system upgrades to virus infections. In addition to a backup of
your data and programs, you need a backup of your bootable operating system so you
can start the PC even when something happens to your hard disk. This is what
emergency boot disks are for.
Typically, a boot disk will be used in the following situations:
• Booting the System: In the event that your hard disk fails
and is unable to boot, you can use your boot disks to start up
the system. This will allow you the opportunity to troubleshoot
and hopefully correct the problem with the hard disk.
Sometimes a hard disk will be accessible after booting from a
floppy disk, even if the hard disk itself will not boot. Without a
boot disk, you are dead in the water since you will be unable to
even start up the system at all.
• Disaster Recovery: Ideally, the restore software for your
backup media should be on an emergency boot disk. This will
enable you to recover from a hard disk disaster and restore
your system from your last backup media set(s).
• Virus Detection and Disinfection: Many viruses, especially
boot sector infectors, will automatically load into memory
every time the hard disk is used to boot the PC. Once in
memory they will interfere with attempts to remove them from
the system. To avoid this, a virus disinfection session should
always be done after a boot from a floppy disk that is known to
be clean and functional.
• Hard Disk Upgrade or Installation: A new hard disk
normally comes unformatted with no operating system on it,
and therefore cannot be used to boot the system. The normal
way to get the operating system onto the disk is to use a boot
floppy to start up the system and then transfer the operating
system files to the hard disk.
I strongly recommend that every PC owner have a set of boot disks for their system,
for use in any of the above circumstances.
Boot Disk Contents
Depending on how you use your system, you may have one or several boot disks. The
main reason for this is that programs are getting larger and larger, and it can be
difficult to get all of the contents that you need onto one disk. In addition, many
programs such as the Norton Utilities or backup software with disaster recovery, will
volunteer to create their own boot disks, sometimes called "emergency disks" or
"rescue disks". These will normally use separate physical disks, and sometimes
duplicate some of the information that is on your "homemade" disks. There is nothing
wrong with this at all, as long as you don't end up with a ridiculous quantity of boot
disks.
Regardless of how many disks you have, you should always have duplicates of every
one in the set. The simple fact is that floppy disks have a high rate of failure, and you
do not want to get a "Sector not found reading drive A:" error as you attempt to reboot
to recover from some disaster.
If you have multiple boot disks, it isn't really necessary for every one of them to be
bootable. ("Huh?") What I mean is that if you only have one boot disk that holds all
the files you need, you must ensure that it contains the operating system files that will
let you boot the system. If your boot disk set includes several disks holding items such
as your antivirus software or other files, they don't all need to be bootable as long as
you have at least two disks that will let you boot the system.
Remember that especially with the operating system and system utilities files, you
need to put on the disk the programs that are appropriate to whatever operating system
you are using. Putting a copy of DOS 6.22's FORMAT.COM on a Windows 95 boot
disk is pointless, because it will not run if the system is booted by that disk (which is
DOS 7.x).
Here are some of the items that I think it is important for you to consider putting on
your set of boot disks:
• Operating System: The boot disk (at least some of them)
must be capable of booting the computer, of course. This
means creating a boot disk that contains whatever files your
operating system needs to boot up; see this section for more
on creating the boot disk itself.
• System Configuration Files: Assuming you are using DOS or
a DOS-based operating system like Windows, it's a good idea
to include your CONFIG.SYS and AUTOEXEC.BAT files on the
boot disk. You may want to rename them however, so that
they aren't automatically used by the floppy disk when it boots
(since these files will contain lines referring to items on your
hard disk that may not work when booting from a floppy).
• Hard Disk Partitioning and Format Utilities: You definitely
need utilities on the boot disk that will allow you to partition
and format your hard disk if you need to. This means, at a very
minimum, including the programs FDISK.EXE, FORMAT.COM
and SYS.COM on the disk. Make sure you use the versions
corresponding to the operating system you put on the disk, or
the programs will give an "Incorrect DOS version" error and will
not run. If you have a copy of Partition Magic or a similar
partitioning utility, creating a floppy disk with this utility on it
can be very useful; follow the directions that come with the
software to create one.
• CD-ROM Driver and MSCDEX: In order to allow the
installation of CD-ROM-based operating systems, you must be
able to access your CD-ROM drive. This requires two pieces of
software: a CD-ROM driver and the file system extension
MSCDEX.EXE that comes with the operating system you are
using. Make sure both are included and that the driver is
loaded in the CONFIG.SYS system file and MSCDEX.EXE in the
AUTOEXEC.BAT system file.
• Diagnostic Utilities: If you have diagnostic utility software on
your PC, including a copy on a disk can be very useful for
troubleshooting problems with your system. At a very
minimum, include SCANDISK.EXE so you can scan for file
system problems if need be, and MSD.EXE (Microsoft
Diagnostics) so you can check your system setup, installed
disks, ports and channels.
• Restore Software: If you are using backup software that
comes with disaster recovery capabilities, the application
should create or allow you to create a floppy disk that contains
the program that will restore your system from backup. Make
sure that you create this disk and include it as part of your
boot disk set.
• Image Information: Image information created from your
hard disk file system should be stored on a floppy disk to allow
you to recover from disk problems.
• Antivirus Software: Good antivirus products will include a
floppy-disk-based version of their software, which will allow
you to boot and run the antivirus scanner directly from the
floppy disk. This is very useful since it will allow you to bypass
any viruses on the hard disk. It isn't always possible to do this
with all antivirus software, however, so you may have to boot
from a clean floppy disk and then run the scanner from the
hard disk (which should still work in most cases but isn't nearly
as good.)
• Editor: Many people don't think of this, but including a simple
editor on the disk like DOS's EDIT.COM will allow you to make
quick changes to configuration files if you need them, which
can be a real life-saver.
• Miscellaneous File Utilities: Any other file utilities you may
have that you find useful should definitely be considered. For
example, a simple file-transfer program can be helpful in
setting up a new PC. I sometimes like to have DEFRAG.EXE on
a boot disk, and MEM.EXE is helpful in diagnosing memory
usage on a new or just-upgraded system. PKUNZIP.EXE comes
in handy for dealing with ZIP files, of course. And DOSKEY.COM
is a helpful addition as well to any boot disk, as it will save you
a lot of typing by letting you recall and reuse previously-typed
commands.
If you are using something like Norton Utilities or similar, a good antivirus product,
and a backup program with disaster recovery, you may find that you will end up with
several different emergency boot disks or "rescue disks". Again, there is nothing
wrong with this at all, since floppies are pretty cheap. The only possible disadvantage
is that you will spend more time keeping them all up to date.
Warning: One final word: if you are using a dynamic disk overlay product like
Disk Manager to allow you access to larger hard disks on a system without
proper native BIOS support, you must ensure that you have a boot disk that contains
the drive overlay software as well. Otherwise, if you boot using a regular boot disk,
you will not be able to access your hard disk. This is very important! Usually the disk
manager program itself will come with a utility or menu item to create a boot disk that
includes the overlay software.
Creating Boot Disks
Creating a boot disk is really a very simple affair, although it depends to some extent
on what operating system you are using and what you want to put on the disk. Bear in
mind that many utilities and backup applications will in fact create their own boot
disks for you. This can be a real time saver, although you often will find other things
that you want on your emergency disks that the software decided not to include. See
this procedure for step-by-step instructions on creating boot disks.
Updating Boot Disks
Since the boot disks that you create as backup for your system contain key
information about your PC, you need to update them whenever your system changes.
For most people, this is not very often. In fact, you probably only need to recreate
your boot disk on an annual basis, when you perform other maintenance tasks that
need to be done once per year.
In addition, you will want to update your boot disks immediately if you make major
changes to your system, such as installing a new operating system. Not keeping your
boot disks up to date can leave you stranded at a time when you really need them.

Using Boot Disks


Using a boot disk is really very simple: you put the boot disk into whichever drive is
the boot device, and reset the PC. A cold reset via the reset button is preferred to using
{Ctrl}+{Alt}+{Delete}, since the latter is just a software reset and will not clear the
contents of memory. You could of course cycle the power as well. Just remember
to shut down the system properly first before resetting.
When the PC reboots, it should load the operating system from the boot disk. Some
PCs will not boot from floppy disks, however. This is usually because floppy disk
booting has been disabled via a BIOS setting. If this is the case, you will have to re-
enable floppy disk booting to use your disk.

You might also like