Personally, I have always kept data and OS separated. But lately I have gone one step further.
On drive C: Windows 2000
Drive D: Page file
Drive E: Temp files
Drives F - J saved for mapping to network shares at work through VPN.
Drive K: DVD-rom
Drives L - X: data and mp3s
I have two Seagate 180GB SCSI Barracuda drives in my system. Can you tell I like to store MP3s? = )
I’m currently building a Linux file server and adding 4- IBM GXP60 60GB drives. I know about the high RMA rate of the 75GXPs, but I have used IBM in the past with no issues.
Just a reiteration to what was said above. Keep the data separate from the OS and you will save yourself many headaches when it comes to reinstalling the OS (regardless of operating system).
The fact that IT departments do it is not an argument for why a home user should. Personally, I see no value in having two PF directories on different drives. Perhaps you can elucidate?
Myself:
In fact, I’ve never personally run into one that did myself. I got this tip from a column in (IIRC) PC Magazine some years ago and have followed it ever since. It seems clear to me that a bug like would be easy to create and could easily slip through testing. Barring real value in doing otherwise, I generally take the default installation path.
Myself:
KellyM:
As an IT professional, this is probably second nature for you. I submit that the average home user trying to do likewise will almost certainly end up with a hodgepodge of some programs installed in C:\PF and some in D:\PF. I’ve heard people put forth the idea of doing it to simplify backing up installed program directories but that’s not especially useful except in the case of those programs who insist on putting data files within their own subdirectory structures.
Myself:
KellyM:
Point taken. This is another tip I picked up some time ago from a magazine column and have been using since. There is an important aspect of this tip that I neglected to mention in the original post. It involves forcing the cluster size on the swap partition to the maximum size possible to increase the efficiency of I/O operations in that partition. In the end, it may come down to a wash given seek time. Given the price of memory these days, the best swap file solution may be just to add enough memory so swapping is seldom if ever necessary.
Myself:
KellyM:
Fragmentation is a problem with any file system, even NTFS, but reformatting is not the solution.
If you re-install Windows, all of the important system files in the root directory will be replaced and backed up if necessary. Yes, you might end up with an extra copy of some old configuration file stuck in the root directory but I consider that a small price.
The big advantage of not wiping the drive is that you can keep a copy of both the old \Windows and \PF directories around for a while. You never know when it’d be handy to have a copy of that old INI file for some application to get back some old settings, or maybe some file that got stuck in the TEMP directory a while back, or maybe an old BOOKMARK.HTM file, and so on. In a pinch, you can even get back your entire old installation if you just did a re-installation of the same version. I’ve needed all of these on at least one occasion and they would have been impossible if simply I’d wiped the disk. If everything’s running fine after a few weeks, then you can confidently delete the old directories.
If you have proper backups, and an IT department behind you, wiping may be a reasonable option. Most personal users I know don’t have anything resembling a proper backup.
BTW, I don’t recommend NTFS for home users because if there is a major system failure there’s no good way to recover files on the drive. With a FAT partition, you can always boot to DOS and get your data back. I’ve personally had this happen twice. A Windows 2000 system had an unknown error and completely refused to recognize a particular FAT32 drive. I pulled the drive and put it in a Win98 machine and it too failed to see it. Finally I booted to a Win98 command prompt and hallelujah, there were my files. I lost all the LFNs but my data was still retrievable. Would this have happened if the drive was using NTFS? Maybe, maybe not, I’ll likely never know, but in the absence of a strict backup regimen, I’ll always recommend FAT.
Myself:
KellyM:
I did a google search and couldn’t find a single reference to FAT rot, at least not in regard to disk drives. I’ve been using PCs for 15 years now and have never come across it so your explanation “the general recognition by most of the people that work with Windows systems on a regular basis” sounds a little provincial. Granted FAT has many shortcomings but rotting is not one of them. Its scale, whether on a floppy or on a multi-gigabyte is irrelevant. In principle it works the same way on both. Files that are being modified when there is a system crash can indeed be truncated, as can happen on any file system, but a file that has been closed will not suddenly disappear or be truncated for no reason. Nor will the OS arbitrarily move clusters around and lose them in the process.
Maybe you’re referring to the cluster slack problem, which is real but likewise is common to all file systems though to a lesser degree than FAT. On a 60GB drive, that’s probably a pretty minor issue.
I did the opposite of double posting. I thought I’d posted a reply here last night, but I guess it never showed up.
Thank you all for your advice. From everything I’ve read here so far, it sounds like I should repartition the 3 gig into one drive, and use it for mp3s or replaceable files like that. On the 60 gig drive, maybe a 30 gig partition for Windows and program files (and the swap file I guess, I only have 32 megs of ram, so the swap file is important)? I compulsively keep backup copies of installation programs, so restoring them isn’t a big deal if I have to format C. I really like the idea of a separate drive for data such as email, documents, and photos so that could be the other 30 gigs of the new drive.
I bought the retail version of the hard drive, so it will have what the webpage calls a “software install utility,” which I hope will be useful in partitioning. If not, I’m not afraid of FDisk, since my data will be safely stored on another computer. I don’t think I can use NTFS; I only have a Windows 98 cd. However, I’m looking forward to moving up to Fat32, because I have so many small files.
How can I tell how big a drive the Bios can handle? I read that Windows 98 can’t handle anything bigger than a 60 gig drive, which is why I chose that size. It hadn’t occurred to me that the bios might have limits. I restarted and found out that my bios is a 1992 (I bought the computer in '96 though) Amibios from American Megatrends, but their webpage didn’t have any useful info or upgrades.
I’ve got a question for KellyM and Artemis: You both seem to agree that it’s a bad idea to create a separate, small partition exclusively for the swap file. Are you talking about a partition on the same (60 Gig) drive? Wouldn’t it be smart for Ariadne to create a small partition on the 3 Gig drive for a swap file? It seems to me like the best use for the old drive.
If you absolutely have to install Windoze, you’ll want separate partitions for the OS and data/MP3’s.
You may want many smaller partitions because it’s a pain in the behind to do the proper maintenance if it’s one huge freakin drive. AND when you install SETI@HOME you will not be able to SCANDISK and DEFRAG without uninstalling SETI@HOME because the stupid thing will access the HDD about 5 minutes into the scan. The scan will then start over - ad infinitum.
Now, if you wanna to the right thing, you’ll install Linux. The installatin should partition your hard drive for optimum performance. Easy.
In principle, that doesn’t sound like a bad idea, especially if the other drive were put on the 2nd IDE controller to reduce contention between the two. In Ariadne’s case however, I suspect that the age of the smaller drive means that it’s probably slower than the new one so any gains would be offset by that.
Ariadne, I doubt that a BIOS from 1992 will be able to handle a 60GB drive. In 1992, a 1GB drive was pretty impressive. Just for comparison on what was available then, I checked back in my own records. I see that I bought a 80MB drive in the summer for '91 for $350, giving me a total 120MB disk space, which was pretty impressive. I’ve never used those software installation utilities that come with a new drive. That may be be able to overcome a BIOS limitation, check the docs that come with it. If not, you may be able to get a BIOS upgrade chip that you can install yourself.
Bear this in mind though. The ROI on upgrades to a machine that old (I’m guessing about Pentium 100) is pretty low when you can get a brand new machine that is far superior to what you have now for about $400 not including monitor. If your budget allows, I’d return the drive and apply the price toward a new machine with a drive that’s 20GB or so. That much space should last quite a while unless you’re doing serious digital video editing.
Regarding partition size, 6-8GB should be more than enough for the the OS & programs partition. Keep the rest for data.
[hi-jack] Nuh-uh! I’ve got SETI@HOME on my PC and I’ve never experienced that. Although it could be because I kill all unneccessary programs before doing Scandisk and Defrag[/hi-jack]
If the old drive is not particularly slow, that might not be a bad use. At least this way you won’t interleave data I/O with swap I/O on the same drive and you won’t be flinging heads all over the place, which is what will happen if you put a separate swap partition on the large drive.
The worst case scenario, by the way, is to put a small swap partition after a large data partition. When allocating space, FAT always selects the free cluster with the lowest cluster number, which means FAT always fills the disk from the low end toward the high end. This means that there is going to be a large gap between the data on a FAT partition and the start of the next partition unless the FAT partition is nearly entirely full. If you have a partition after a large FAT partition, getting to that partition will require a large seek. If this partition is heavily used (such as a swap partition), you will see a diminution of performance due to a high proportion of time being spent seeking the drive heads. NTFS does not suffer from this problem; NTFS seeks to balance block allocation across the entire partition in order to minimize average seek times. (NTFS is also designed to avoid fragmentation, which is why it is not as necessary to run defragmenters on a regular basis on NTFS systems. Another reason to use NTFS.)
I would theoretically advise also insuring that the second drive is on the other I/O channel, but that’s not really important as very few IDE controllers actually have independent channels (to save money, the controllers serialize the channels, so even though the IDE specification permits simultaneous operations on both channels, the card is not actually capable of doing so).
Ariadne, wehn you get a HD it should come with software to get around your bios limitations. You can also download that software from the manf web site for free. Thats the place to look for info on what to do, the web site that is.
Set up the old drive as a slave (when you’ve got 60 GB, an additional 3GB isn’t going to mean much, even if you have thousands of MP3s). Create your C partition to be no larger than the old drive (3.2 GB). Use the rest of the drive for your D partition (Win 98 may force the second drive to take the “D” designation; no biggie if it does). For a normal PC user I don’t think that any more than two partitions are justified unless you’re using multiple OS’es, which I assume you’re not. All software except for the OS should be installed on the D partition.
After you have everything installed to your satisfaction, reformat your old drive and use the xcopy/s (or is it xcopy32 with Win98?) command to copy everything from your C partition to your old drive (E?). You may have to boot to a command prompt to do this. Note: this will not work with Win2K or XP, and may not work with ME. To do it with those OSes, you’ll need a partition backup utility like Norton Ghost). Every time you make a major change to your OS (like installing more software), make a habit of doing that copy again (delete all the old files on the old drive before you do this). This will give you a backup if/when your C partition gets trashed; if that happens, boot to a floppy and copy all your files back.
Frogstein, good idea with the OS backup but but XCOPY is not a good way to do it. Nor is any other routine way of copying files like drag-n-dropping them in explorer. The reason is that even though Win9x programs handle long file names, they don’t always use them internally. When storing path/file names in the registry, some of them use the equivalent DOS 8.3 file names instead. The result is that copying those files with a normal mechanism can put the registry out of sync with the actual file names. Why they do this I have not a clue but it can cause a real nightmare for the person who attempts the perfectly reasonable action of making a backup copy of their system files.
As you mentioned, ghost is a good choice if you have access to a copy.
I will defer to you. The last time I tried this I was successful, but I don’t recall if it was Win95 or Win98. I got the idea from a Win 95 FAQ from several years back.
It will work in most cases for the toy versions of Windows (W95, W98, and Me). It will not for NT4, 2000 (NT5) or XP (NT6) because of the requirement of a unique SID and (possibly also) because XCOPY does not respect NTFS ownership and permissions (use SCOPY for that).
There are situations where it won’t work for the toy versions of Windows, but those situations should be uncommon.
Here’s an excerpt chosen at random from my own system’s registry. It shows the file that’s used to open CPP (C++) files. This registry entry was created by Microsoft Visual Studio.
The long filename (LFN) version of this would be
The “MICROS~2” portion of the name is caused by the fact that this subdirectory was the second one created that began with the letters “MICROS”. The first one that was created uses “MICROS~1” as its short name and so on for others with similar names. The same rules apply to similar file names with a particular directory.
If I use XCOPY to make a backup of my \Program Files tree, and then later try to copy that same directory back to the C: drive, those two directores may or may not get copied back in the same order as the order in which they were created originally. If they get copied in the wrong order, the other “Microsoft…” directory will get assigned the short name of MICROS~2. Any programs that use those registry references will be looking in the wrong place for data. In other words, part of the system will be utterly hosed.
If someone’s system is set up so that there is not a single instance of this situation in the \Windows or \Program Files trees, then XCOPY or drag-n-drop will work just fine because no number higher than 1 will ever be used. If not, there’s potential for a situation that could cause errors that are extremely difficult to track down.
It will please KellyM no end that I did a cursory check in Windows 2000 and it seems to use the long file names exclusively, which is what the other Windows variants should do.
