Should I Partition the Hard Drive

[Guest]

Is there any reason to partition a hard drive drive in Windows XP? Do I gain
speed, efficiency, space? The drive is 300GB's and I'll slipstream SP2 into
an install disk so XP will recognize the full 300 GB's.

 

[Jerry]

Not if you format the whole thing in NTFS.

 

[Pegasus \(MVP\)]

Is there any reason to partition a hard drive drive in Windows XP? Do I gain
speed, efficiency, space? The drive is 300GB's and I'll slipstream SP2 into
an install disk so XP will recognize the full 300 GB's.

I always partition my client's PCs. There are several reasons:
- I can format drive C: and re-install Windows without
loss of data, as all data is kept on drive D:.
- I can use imaging products such as DriveImage or TrueImage
to save my installation, then restore it when something
goes wrong.
- There is an intrinsic division between OS and applications
on the one hand (drive C, and user data on the other
hand. Without partitioning, users will save their files
just about anywhere, making the backup process much
less dependable.

However, as Jerry said, you do not gain anything in terms of
speed, efficiency or space. In fact you lose some flexibility when
you partition a disk, because if you partition it incorrectly then
you could run out of space on drive C: while having lots of
free space on drive D:.

 

[Ron Martell]

Is there any reason to partition a hard drive drive in Windows XP? Do I gain
speed, efficiency, space? The drive is 300GB's and I'll slipstream SP2 into
an install disk so XP will recognize the full 300 GB's.

There is no "right answer" to this question.

Most experienced users will have different answers to it, at least in
the details.

Think of your hard drive as a filing cabinet, because that is what it
really is - a storage place where you want to keep things neat and
organized.

Having the whole drive as one large partition is like having one
massive filing drawer for all of your files. It can work, but it
takes a bit of housekeeping and self-discipline to ensure that things
can always be found.

Partitioning the drive is the equivalent of replacing the one huge
file drawer with several smaller drawers with the same total capacity.
This will usually be easier to keep tidy and organized, at least for
most people.

One danger in partitioning a drive is that you may not get the
partition sizes right, particularly if you split the drive a number of
partitions for different purposes. What too often happens is that the
partition used for music files, for example, will prove not to be
large enough and there will be scads of unused space in some of the
other partitions. And adjusting partition sizes without destroying
the data can be a daunting task. It can be done, but it requires a
third party partitioning program such as Partition Magic or BootItNG
and can be quite time consuming.

MVP Jim Eshelman has a web page where he discusses partitioning issues
at http://www.aumha.org/a/parts.htm

Good luck


Ron Martell Duncan B.C. Canada
--
Microsoft MVP
On-Line Help Computer Service
http://onlinehelp.bc.ca

"The reason computer chips are so small is computers don't eat much."

 

[Timothy Daniels]

I always partition my client's PCs. There are several reasons:
- I can format drive C: and re-install Windows without
loss of data, as all data is kept on drive D:.
- I can use imaging products such as DriveImage or TrueImage
to save my installation, then restore it when something
goes wrong.

You can also save a partition that contains the OS
AND the data and restore them both when something
goes wrong. It seems you're assuming that the something
that goes wrong will always be part of the OS when, in
fact, *any* file or data structure can get corrupted

- There is an intrinsic division between OS and applications
on the one hand (drive C, and user data on the other
hand. Without partitioning, users will save their files
just about anywhere, making the backup process much
less dependable.

The problem is that there are lots of files that aren't
clearly system files or user files, such as boot.ini,
dialup connectoids, 3rd party utilities, Master Boot
Record, favorites, ISP servers names and account
user names and passwords in Outlook Express, etc.
What do you back up and what do you not back up
as part of the "OS"?

*TimDaniels*

 

[Hans-Georg Michna]

Is there any reason to partition a hard drive drive in Windows XP?

No, normally not.

Do I gain speed, efficiency, space?

No, instead you lose some speed and you lose flexibility.

The drive is 300GB's and I'll slipstream SP2 into
an install disk so XP will recognize the full 300 GB's.

Yes, that's the best way to go about things.

That said, there are some reasons to use partitions. One of them
is that you want to install more than one operating system.

Some people also give false reasons for partitioning, usually
revolving about backup, safety, cleanliness or more abstract
concepts, but these are almost always irrational.

Hans-Georg

 

[Pegasus \(MVP\)]

See below.

You can also save a partition that contains the OS
AND the data and restore them both when something
goes wrong. It seems you're assuming that the something
that goes wrong will always be part of the OS when, in
fact, *any* file or data structure can get corrupted

Imaging a drive that contains data is in most cases impractical,
because of size limitations.

Corrupted data files can easily be restored from backup.
A corrupted OS is much harder to restore, because you
often do not know what's gone wrong. The countless
posts in these newsgroups are a clear illustration.

The problem is that there are lots of files that aren't
clearly system files or user files, such as boot.ini,
dialup connectoids, 3rd party utilities, Master Boot
Record, favorites, ISP servers names and account
user names and passwords in Outlook Express, etc.
What do you back up and what do you not back up
as part of the "OS"?

*TimDaniels*

When you take an image of drive C: then you're backing
up all of the above. The two backups should complement
each other: Back up data files weekly, take an image snapshot
prior to every major change but at least twice a year.

 

[Pegasus \(MVP\)]

No, normally not.


No, instead you lose some speed and you lose flexibility.


Yes, that's the best way to go about things.

That said, there are some reasons to use partitions. One of them
is that you want to install more than one operating system.

Some people also give false reasons for partitioning, usually
revolving about backup, safety, cleanliness or more abstract
concepts, but these are almost always irrational.

Hans-Georg

Six months ago I had one of your irrational moments. One
of my client's servers went AWOL during a minor upgrade.
The client needed the machine back within one hour. A
parallel installation of Windows, followed by a full restoration
from tape, would have taken at least half a day. The restoration
from an image file restored the system in 45 minutes.

By the way, would you care to explain why a system with
a partitioned disk should lose some speed?

 

[Bruce Chambers]

Is there any reason to partition a hard drive drive in Windows XP? Do I gain
speed, efficiency, space? The drive is 300GB's and I'll slipstream SP2 into
an install disk so XP will recognize the full 300 GB's.

Placing data files on a partition or physical hard drive separate from
the operating system and applications can greatly simplify system
repairs/recoveries and data back-up.


--

Bruce Chambers

Help us help you:



You can have peace. Or you can have freedom. Don't ever count on having
both at once. - RAH

 

[Guest]

Placing data files on a partition or physical hard drive separate
from
the operating system and applications can greatly simplify system
repairs/recoveries and data back-up.

I agree. I try to keep user-created files off of C: - this allows for
backing up C: as an image/whatever, which can be restored without having
to reinstall Windows. If D: just has data files on it, it can be backed
up/restored via simple file-copy type of operations, which allows for
selective/incremental backup and versioning.
I also try to make all programs default to working on the D: drive as
default, instead of C: (including CMD windows). This encourages the user
to keep their files there, and helps limit damage to the C: drive when
the user does something dumb or a program goes nuts or whatever.

I find that even machines with huge numbers of application programs need
more than 15-20Gb for the C: drive.
You can move many things like "My Documents", TEMP, Temporary Internet
Files, etc. to another drive. You could even install some applications
onto a different drive than C:, especially ones that are easy to
reinstall or include huge data files (games come to mind).
Some applications insist on keeping data on the C: drive and nowhere
else, but with a bit of effort you can even get things like Outlook to
keep most of its data off of C:.

Most of the arguments that I can think of for partitioning the drive are
related to backup - and more importantly, restore. Backup is easy - the
thing to consider more is restoration. When considering how to backup, I
first decide what I want to do after disaster strikes, and then set up to
be able to do that. The disaster can be a hardware failure, data
corruption, viruses, or user errors (like accidental deletion) - it may
involve single files, or an entire partition/drive.

I can think of few downsides to a bit of partitioning. One is possibly
wasting some space and/or running out of room on one partition. By
choosing the partitions sizes, and possibly moving files from one drive
to another, one can usually limit this to a small fraction of the drive
size. Performance/speed should not be affected significantly, and some
factors could favor partitioning here. For example, it may be easier to
limit fragmentation or defragment if the system files and data files are
kept separate.

 

[Timothy Daniels]

[....] If D: just has data files on it, it can be backed
up/restored via simple file-copy type of operations,
which allows for selective/incremental backup and
versioning.

That's the first convincing argument (for me)
that I've heard.

[....] You can move many things like "My Documents", TEMP,
Temporary Internet Files, etc. to another drive. You could
even install some applications onto a different drive than C:,
especially ones that are easy to reinstall or include huge
data files (games come to mind).
Some applications insist on keeping data on the C: drive
and nowhere else, but with a bit of effort you can even get
things like Outlook to keep most of its data off of C:.

Do you have any info or links for putting Outlook's and
OE's data (such as email, saved posts, newsgroup cache)
on drive D:?

*TimDaniels*

 

[Ron Martell]

Six months ago I had one of your irrational moments. One
of my client's servers went AWOL during a minor upgrade.
The client needed the machine back within one hour. A
parallel installation of Windows, followed by a full restoration
from tape, would have taken at least half a day. The restoration
from an image file restored the system in 45 minutes.

By the way, would you care to explain why a system with
a partitioned disk should lose some speed?

The reasoning behind this is that with a partitioned hard drive there
will be longer travel distances for the drive head mechanism, thereby
increasing the seek time.

This presumes that the partitions are not completely filled, which is
most often the case.

Let's assume a 120 gb drive with 4 partitions.

1. Operating system 10 gb allocated 6 gb used
2. Application programs 30 gb allocated 10 gb used
3. User data files 20 gb allocated 5 gb used.
4. Downloads 40 gb allocated 10 gb used.

If the drive were all in one partition there would be 31 gb of the 100
gb used, or 31%. Allowing for a bit of unused space scattered
throughout the drive the longest distance the heads would have to
travel should not exceed 40% of the maximum.

With the partitioning scheme as per my example the longest seek
distance would be from the beginning of the first partition to the end
of the data content in the fourth partition, covering about 60 gb of
the drive, or 60% of the maximum.

Hope this explains the situation.




Ron Martell Duncan B.C. Canada
--
Microsoft MVP
On-Line Help Computer Service
http://onlinehelp.bc.ca

"The reason computer chips are so small is computers don't eat much."

 

[Pegasus \(MVP\)]

The reasoning behind this is that with a partitioned hard drive there
will be longer travel distances for the drive head mechanism, thereby
increasing the seek time.

This presumes that the partitions are not completely filled, which is
most often the case.

Let's assume a 120 gb drive with 4 partitions.

1. Operating system 10 gb allocated 6 gb used
2. Application programs 30 gb allocated 10 gb used
3. User data files 20 gb allocated 5 gb used.
4. Downloads 40 gb allocated 10 gb used.

If the drive were all in one partition there would be 31 gb of the 100
gb used, or 31%. Allowing for a bit of unused space scattered
throughout the drive the longest distance the heads would have to
travel should not exceed 40% of the maximum.

With the partitioning scheme as per my example the longest seek
distance would be from the beginning of the first partition to the end
of the data content in the fourth partition, covering about 60 gb of
the drive, or 60% of the maximum.

Hope this explains the situation.

Your suggested mechanism would be valid if the machine kept
hopping from drive C: to drive D: and back again. It would not
apply in those cases where whole batches of files are read from
drive C: (e.g. during the start-up phase) or from drive D: (e.g. while
processing user files).

Regardless of this, I wonder if the argument isn't largely academic.
In other words, are you in a position to back up your claim with
actual measurements carried out in a controlled environment?
I have a faint suspicion that if this effect exists then it is only
detectable with a stopwatch. Users wouldn't notice the slightest
difference. I'll wait with interest for your figures, with an
indication of the source (or a description of your test configuration).

 

[Uncle John]

Hi All

This is an interesting thread for with experience disaster recovery
specialist with Windows applications since Windows version2 and before then,
with DOS. The pro's and con's of partiton are both well put forward. I
suggest the choice should depend on how many disks there are on the system,
bearing in mind it is easier to restore a whole system including data than
bits and pieces afer a disaster.

If the user has only one very large hard drive it is much better to split it
into two partitions for backup simplification. If there are two or more
similar sized drives then do not partiton either. Recovery from disaster
from drive is by imaging the system drive to the backup drive with System
Restore only enabled for the system drive [C:]. A good imaging application,
like Acronis True image or Drive Image/Ghost 2004 is used to backup the
WHOLE of the system drive to the backup drive. My experience of disasters is
that users normaly cannot say what happened and it is simplest to go
straight to restoration with a Bootable CD which allows selection of the
backed up image on drive d: and restoration to drive c: Do not put anything
else on the backup drive it may interfere with the stored images. [However
Acronis will create a safe recovery zone on the backup drive which is not
affected by Windows applications but which can be mounted as a virtual drive
for partial restorations in Windows; in this case it is safe to put
temporary files and internet files on Drive d;]
What do you think?

BTW I suggest that it is safer to backup data it to removable media (tape or
opltical disk) which can be archived in a safe place.

 

[Pegasus \(MVP\)]

Hi All

This is an interesting thread for with experience disaster recovery
specialist with Windows applications since Windows version2 and before then,
with DOS. The pro's and con's of partiton are both well put forward. I
suggest the choice should depend on how many disks there are on the system,
bearing in mind it is easier to restore a whole system including data than
bits and pieces afer a disaster.

If the user has only one very large hard drive it is much better to split it
into two partitions for backup simplification. If there are two or more
similar sized drives then do not partiton either. Recovery from disaster
from drive is by imaging the system drive to the backup drive with System
Restore only enabled for the system drive [C:]. A good imaging application,
like Acronis True image or Drive Image/Ghost 2004 is used to backup the
WHOLE of the system drive to the backup drive. My experience of disasters is
that users normaly cannot say what happened and it is simplest to go
straight to restoration with a Bootable CD which allows selection of the
backed up image on drive d: and restoration to drive c: Do not put anything
else on the backup drive it may interfere with the stored images. [However
Acronis will create a safe recovery zone on the backup drive which is not
affected by Windows applications but which can be mounted as a virtual drive
for partial restorations in Windows; in this case it is safe to put
temporary files and internet files on Drive d;]
What do you think?

BTW I suggest that it is safer to backup data it to removable media (tape or
opltical disk) which can be archived in a safe place.

Not only is it safer to backup data to a removable media but it is
actually mandatory. Here are the main causes of loss of data:
1. User error
2. File corruption
3. Media corruption
4. Theft
5. Fire, water damage

If the backup medium (e.g. a hard disk) is kept in the PC then
it is likely to be affected by all of the above. When deciding if
your current storage place for the backup medium is safe,
consider what will happen under 4. and 5 above.

Incidentally, a "fire-proof" safe is useless for backup media.
Your paper documents may survive, even when burnt to
a crisp brown colour, but your disks/tapes won't. You
can easily test this in your own oven!

 

[Ron Martell]

Your suggested mechanism would be valid if the machine kept
hopping from drive C: to drive D: and back again. It would not
apply in those cases where whole batches of files are read from
drive C: (e.g. during the start-up phase) or from drive D: (e.g. while
processing user files).

Startup would be loading soleleyfrom the operating system partition if
only the operating system was loaded at startup. If application
programs, such as antivirus and/or various background tools and toys
were being loaded then this would require access to the application
programs partition as well. And if any of these application programs
had a data file associated with it (e.g. a P.I.M. program) then that
would then bring the data files partition into use.

And when an application program is launched it will invariably also
require files from the operating system (.DLLs from the \system32
folder for instance) and very likely a data file as well.

Regardless of this, I wonder if the argument isn't largely academic.
In other words, are you in a position to back up your claim with
actual measurements carried out in a controlled environment?
I have a faint suspicion that if this effect exists then it is only
detectable with a stopwatch. Users wouldn't notice the slightest
difference. I'll wait with interest for your figures, with an
indication of the source (or a description of your test configuration).

The following specs are for a Western Digital Caviar 250 gb 7200 rpm
drive:

Read Seek Time (Average) 8.9 ms
Write Seek Time (Average) 10.9 ms (average)
Track-To-Track Seek Time 2.0 ms (average)
Full Stroke Seek 21.0 ms (average)

http://www.westerndigital.com/en/products/Products.asp?DriveID=41


Ron Martell Duncan B.C. Canada
--
Microsoft MVP
On-Line Help Computer Service
http://onlinehelp.bc.ca

"The reason computer chips are so small is computers don't eat much."

 

[Pegasus \(MVP\)]

Startup would be loading soleleyfrom the operating system partition if
only the operating system was loaded at startup. If application
programs, such as antivirus and/or various background tools and toys
were being loaded then this would require access to the application
programs partition as well. And if any of these application programs
had a data file associated with it (e.g. a P.I.M. program) then that
would then bring the data files partition into use.

And when an application program is launched it will invariably also
require files from the operating system (.DLLs from the \system32
folder for instance) and very likely a data file as well.




The following specs are for a Western Digital Caviar 250 gb 7200 rpm
drive:

Read Seek Time (Average) 8.9 ms
Write Seek Time (Average) 10.9 ms (average)
Track-To-Track Seek Time 2.0 ms (average)
Full Stroke Seek 21.0 ms (average)

http://www.westerndigital.com/en/products/Products.asp?DriveID=41

I think we both know that raw technical data does not mean
much when it comes to measuring actual performance in a
complex environment. You might argue, for example, that it's
much faster to travel by car from A to B in downtown LA,
because a car travels much faster than a pushbike. The actual
travel time depends, of course, on a great many factors, and
a bike could easily be as fast as a car.

Seeing that you did not offer any hard figures about real-life
performance of a partitioned vs. an unpartitioned drive
(other than raw access figures), I conclude that your
claim to better performance of an unpartitioned drive is
largely based on theoretical reasoning, without any hard
evidence to back it up.

 

[Hans-Georg Michna]

Six months ago I had one of your irrational moments. One
of my client's servers went AWOL during a minor upgrade.
The client needed the machine back within one hour. A
parallel installation of Windows, followed by a full restoration
from tape, would have taken at least half a day. The restoration
from an image file restored the system in 45 minutes.

Pegasus,

you don't need partitions for a partial restore.

But I won't argue with that. When somebody has a really good
reason to use partitions, then he should use them.

By the way, would you care to explain why a system with
a partitioned disk should lose some speed?

Ron already explained it. The reason is longer head travel.

Of course, this depends on many factors.

A negative example would be an extra partition for the swap
file.

In one partition a good defragmenter like Raxco's Perfect Disk
can optimize file locations according to usage. Even Windows
itself does this to some extent.

Hans-Georg

 

[Hans-Georg Michna]

I think we both know that raw technical data does not mean
much when it comes to measuring actual performance in a
complex environment. You might argue, for example, that it's
much faster to travel by car from A to B in downtown LA,
because a car travels much faster than a pushbike. The actual
travel time depends, of course, on a great many factors, and
a bike could easily be as fast as a car.

Seeing that you did not offer any hard figures about real-life
performance of a partitioned vs. an unpartitioned drive
(other than raw access figures), I conclude that your
claim to better performance of an unpartitioned drive is
largely based on theoretical reasoning, without any hard
evidence to back it up.

Pegasus,

as far as I'm concerned, this is true, but the reasoning is
fairly sound.

On the other hand, the positive effects of disk optimization are
often overestimated (as I myself argue in
http://www.michna.com/kb/WxDefrag.htm in a related context).

To give a simple example, if a computer has a lot of RAM, then
disk performance can play a much smaller role, because a lot of
disk data can be in the RAM cache.

But then I'm only giving a general recommendation from my
personal point of view, while trying to be rational. I carefully
mentioned that there can be good reasons to have more than one
partition. And that there are also a couple of bad reasons
around. I hope that this is good enough to help all readers to
make an educated decision.

After this discussion the readers will understand the points
even better.

Hans-Georg

 

[Pegasus \(MVP\)]

Pegasus,

as far as I'm concerned, this is true, but the reasoning is
fairly sound.

On the other hand, the positive effects of disk optimization are
often overestimated (as I myself argue in
http://www.michna.com/kb/WxDefrag.htm in a related context).

To give a simple example, if a computer has a lot of RAM, then
disk performance can play a much smaller role, because a lot of
disk data can be in the RAM cache.

But then I'm only giving a general recommendation from my
personal point of view, while trying to be rational. I carefully
mentioned that there can be good reasons to have more than one
partition. And that there are also a couple of bad reasons
around. I hope that this is good enough to help all readers to
make an educated decision.

After this discussion the readers will understand the points
even better.

Hans-Georg

The reasons for having a single partition in order to "improve"
performance may be sound but ***not a single respondent***
in this long thread has been able to back up his claim with a
test process than can be duplicated by anyone else. Until such
a test process is posted (and duplicated), the claim remains
what it is: an unproven hypothesis.