Partitioning a new HDD

[Alex-ander]

I am installing a new HDD that i just built. What's the best way to partition the HDD so i have a C and D partition.

I will be installing windows XP on it for now.

[Smokey]

When you put you windows CD in to install XP you have the choice to delete or create partitions, just create two partitions to the size of your choice and then choose which one you want to install xp on. The sizes are in MB so make sure you make them big enough, 70000MB = 70GB.
When XP is installed you may need to navigate to DISK MGR to format your 2nd partition!

[Taiji]

I'd suggest you make 3 partitions. I want 1 for OS, one for programs and one for data. Basically I don't want to be downloading stuff to my OS or programs drive because of the fragmentation that results. But I'm very sensitive to performance differences, and if you're not then maybe it doesn't matter.

[Simetrical]

With NTFS you'd better be defragmenting regularly anyway, so defragmentation shouldn't be an issue. If you aren't defragmenting, you may well cause more fragmentation by using more partitions, not less, since you'll have less free space in each partition (and fragmentation mainly occurs when there's not enough free space).

[Taiji]

With 3 partitions the user just has 2 partitions that will ever need defragmentation and they are smaller, I don't see any superior option.

2 parts, by comparision, encourages users to download to the apps or OS drive which means they will need to defrag more often, and it will take more time when they do so.

And actually, I think it's more accurate to say that fragmentation mainly occurs when a file is altered and no longer fits in the blocks allocated to it.

[Simetrical]

With 3 partitions the user just has 2 partitions that will ever need defragmentation

The OS and programs partitions don't fragment over time? They'd fragment less, certainly, but you really don't ever need to defragment them?

2 parts, by comparision, encourages users to download to the apps or OS drive which means they will need to defrag more often, and it will take more time when they do so.

That's why you do it overnight, or in the background.

And actually, I think it's more accurate to say that fragmentation mainly occurs when a file is altered and no longer fits in the blocks allocated to it.

Which tends to happen a lot more often when there's less free space, because then the OS can't leave files as much room to grow into. If a partition is 90% full, it will fragment much faster than if 60 or 70% full.

Personally, I prefer to put everything on one partition if possible. It just makes things simpler. If you want to keep your files when reinstalling the OS, you can always copy them.

[Taiji]

The OS and programs partitions don't fragment over time? They'd fragment less, certainly, but you really don't ever need to defragment them?

If one hardly ever needs defragging then you can't refer to it in the plural. So I think you misread.

That's why you do it overnight, or in the background.

No mate, you do it as little as possible. You don't get all unefficient and just say "I'll leave it on overnight because I wanted 2 partitions instead of 3". You just do as I say and defrag less often.

Which tends to happen a lot more often when there's less free space, because then the OS can't leave files as much room to grow into. If a partition is 90% full, it will fragment much faster than if 60 or 70% full.

So you agree that freespace is not a factor at all unless what I described is taking place. That's good. Size is mostly irrelevant to the question of how many partitions you make. It's relevant to the size of each partition, but not to how many. Of course if you have a drive so small that you cannot split 3 usable partitions, like some SSDs and older HDDs, then you don't use it like a conventional drive.

Personally, I prefer to put everything on one partition if possible. It just makes things simpler. If you want to keep your files when reinstalling the OS, you can always copy them.

Personally I prefer to defrag as little as possible. So the opposite to what you suggest there. I can appreciate that it looks more simple to do it with 1 part, but it sounds like a waste of time to me.

I have a 40 for XP, an 80 for 7, 120 for app data and another 2 120s for data. I probably need to defrag my OS parts 3 times a year, my apps drive 2 times and my data .... probably never. Well I say probably never but note that I split it in 2 so that I only defrag 120 if a data part starts underperforming.

So with a 500gb drive I typically defrag a maximum of 600gb a year. If I had 1 part that would grow to, well you should be able to imagine; many TBs.

It just takes a bit of effort and organisation to be as lazy as possible

[Simetrical]

No mate, you do it as little as possible. You don't get all unefficient and just say "I'll leave it on overnight because I wanted 2 partitions instead of 3". You just do as I say and defrag less often.

It's not inefficient to have your computer do things automatically in the background for you. That's what computers are supposed to do.

So you agree that freespace is not a factor at all unless what I described is taking place. That's good.

If you have enough free space, you'll see negligible fragmentation. If you have too little, you'll see tons of fragmentation. (At least, typically. You can come up with theoretical counterexamples either way.) It also depends on access patterns, of course: lots of files being created and then some percentage being deleted all the time will cause a lot more fragmentation. But even a system partition will quickly become fragmented in normal use if it becomes too full, e.g., 95%. And if you're leaving it only 80% full, then you could be more efficiently sharing that free space with your other partitions. Although of course, disk space is so cheap today that it makes no big difference unless you have an SSD, in which case you hardly care about fragmentation.

Personally I prefer to defrag as little as possible. So the opposite to what you suggest there. I can appreciate that it looks more simple to do it with 1 part, but it sounds like a waste of time to me.

There's no time wasted if you set up the defragmentation to happen automatically. On the other hand, if you make one of your partitions too small and need to reclaim space from another partition, that's a waste of time, because you have to shut down the system and have it move all the files around offline. Fragmentation can be done in the background.

. . . Of course, this is all academic to me. I mainly use Linux, which tends not to fragment (and indeed, doesn't currently support defragmentation in most cases). And my Windows machine has an SSD, so fragmentation is irrelevant. I used to use only Windows, though, back in 2006 or so . . .

[Taiji]

Suggesting that I refered to background use of defrag as inefficient compared to ... to .... what? Oh, there was no comparison on those grounds! ...Please pay attention!

I am saying that 2 parts is less efficient than 3. It doesn't matter when you are inefficient, in the background or overnight, you are still inefficient by comparison. My method is simply more efficient than yours, it should really be obvious, I must be making some mistake in explaining it...

The point you make about a system drive with 95% full is flawed - if your 'system' drive is that full then you are doing something seriously wrong. Either you did not set up a reasonable size or you installed more than your OS, drivers and small, frequent use apps, like winrar, skype, etc. The data drive/s can get full, and the apps drive too when things get tight, but never the system drive... or why call it that?

Anyway, you seem to be saying that even if you do realise that I am right it won't make any difference for you. So there is no point going any further trying to help you out by convincing you.

I bid you a good day and all the best

[Simetrical]

I am saying that 2 parts is less efficient than 3. It doesn't matter when you are inefficient, in the background or overnight, you are still inefficient by comparison.

Inefficiency that isn't visible to the user is academic. Defragmentation that happens in the background at idle priority has zero cost to the user, and so it doesn't matter even if you have to defragment ten times as much. Yes, it's less "efficient" in that it's using your disks more to achieve broadly the same effect, but there's no actual cost to using your disks more per se, so who cares?

Anyway, you seem to be saying that even if you do realise that I am right it won't make any difference for you.

Well, no, since I use ext4 on Linux, which tends not to fragment. I mostly don't discuss tech things here to get practical advice, more because I find the subject matter interesting.