View Full Version : proper partition strategy
kmlevine
04-28-2004, 09:48 PM
I have 2 HDD, an 74 Gig (10k rpm) and a 160 Gig (7.5k rpm) , neither have partitions. I use my XP system for working on digital imags in Photoshop and other programs as well as converting old vinyl LPs to wav files. I’m seeking recommendations for a partition strategy; if I need a separate partition for page files. programs, data...I'm a newbie at hard disk management. Thank you in advance.
Sylvander
04-29-2004, 05:22 PM
1. Keep the C: partition as small as possible but big enough to hold the Windows folder and the Programs folder plus odds and ends.
I'm using 1.20 Gb of 1.40 GB for Win98, you may need 2 or 4 times that for XP, I don't know.
Use small clusters. I use "Disk Frontier" to study file sizes. There are a LOT of files of less than 1,000 Bytes.
Keep it "lean & mean" and back it up regularly when clean.
Reformat & restore a clean backup to eliminate software problems.
REHOME
[preferably on another physical drive, but another partition if not]
a. My Documents.
b. E-mails for all identities.
c. Internet Explorer Favourites.
d. Temporary Internet Files.
e. Re-home the [Win98] Microsoft Address Book as shown here http://tinyurl.com/2adxv . Use the key “HKEY_CURRENT_USER\Software\Microsoft\WAB\WAB4\Wab FileName” to specify its new address. [Its normal home address [in Win98] is C:\WINDOWS\Application Data\Microsoft\Address Book.]
When you re-format and re-store, these will still be up-to-the-minute.
2. Make different partitions to hold files with different characteristics.
Huge files [like wave or MP3] should be put in a partition big enough to hold them all. Use the biggest cluster sizes. I have 30 - 100 MB wave files & 4 - 8 MB MP3's. Study your file sizes and use suitable cluster sizes to minimize "slack". You will then find that defragmenting is quicker, backup is quicker, restoring is quicker.
You backup different file types and therefore partitions at different frequency or intervals.
Files that change frequently [like on the C: partition] should be backed up frequently.
Files that don't change often, or are not so critical don't need backed up so often.
3. SEPARATE PHYSICAL DRIVE
I keep my Windows installation files off the C: drive and on this. Likewise with the fixed-size swap file.
I keep the installation files for every program I ever have [or are likely to be] installed, on CD-RW disks, but you could keep them here.
Every one is in its own folder [the name identifies it] with a wordpad file saying where & when it was got and what it does and how [text copied from the web page].
Fruss Tray Ted
04-29-2004, 06:12 PM
Other than Sylvander's (sorry bro ;) ) C: drive being a little too small, even the suggested size (esp if you have lot's of programs), I would add a permanent Swap File partition to his list. I think he forgot to list this.
Primary partition sizes will vary with your operating system. Where 1 gig is plenty for Windows 95, 2 or slightly more would be better for 98. My next 98 setup I plan on 2.5 gigs. Granted, with 700 megs for C; right now, defrag is fast, security scans are a breeze and the 800mhz Duron pc is comparable to a 1.2 gig or greater Athalon that uses a whole harddrive as C: It is definitely faster than a friend's Dell with a 1.2 gig but I don't know how much RAM. I think they are close at 384 megs or so..
Where you mention XP, if you offload all functions that you can (as explained above), I would use around 5-7 gigs. I know other regulars here prefer closer to 15g but it really depends on how many apps you plan on having available. Remember to disable them in startup (msconfig or similar) if you want the pc to run as fast as it can without the bloat of apps running in the background.
Sylvander
04-29-2004, 06:32 PM
I resized my C: partition by [copying & pasting data files to other partitions] backing up everything left on it, re-partitioning to a better size, then restoring the backup.
I made it that size because it's all it needs.
If it begins to run out of spare room I'll make it bigger [it's part of a 10 GB HDD].
I mentioned the "fixed-size swap file" in note 3.
It's not alone on the partition, but it is on a separate physical HDD.
I figure it could be reading from the swap file [on the D: drive] at the same instant as writing to the C: drive.
Except that I have them both on the same [primary] controller.
Hey ho :D
Fruss Tray Ted
04-29-2004, 06:39 PM
Sorry, musta missed it or had a bout of 'selective reading', y'nno, sorta like selective hearing... :D
Sylvander
04-29-2004, 06:55 PM
You are not alone in that. :)
Now, what's my name again?
Ah yes, Sylvander.
Fruss Tray Ted
04-29-2004, 07:50 PM
Yunno something else? That pink sign you got has me thinking of a song about Timothy Leary. I think it's time to hear some music by the Moody Blues... Something about being 'On the outside, Looking IN!,,, :(
Paul Komski
05-01-2004, 04:26 AM
It is very hard to suggest a good partitioning strategy that suits everyone. So the best advice I can think of is to get used to using good partitioning tools (Partition Magic and Drive Image are my own pair of must haves) and then you can be flexible, change your own set-up to suit your (probably ever-changing) needs and quickly make and restore both hdd and off-the-pc backups.
If you are running WinXP on modern hardware with lots of RAM it is my growing belief that one can forget about all the minor tweaking that goes on to try and squeeze the last ounce of performance out of one's pc. The same goes for tweaking the paging file (which can become insignificant in importance for non-techies if one has loads of ram) or the cluster size (which is far less critical if one uses ntfs instead of fat). The cluster size only really defines the possible size of the total partition for any given file system - the redundancy of lost or wasted space is of much less important with today's enormous storage capacity.
I agree with keeping the OS partition lean and efficient and if you store the majority of your programs and data on other partitions there is probably no need to have it larger than 3 or 4 GB for WinXP. This makes disk maintenance (chkdsk, defrag, etc) and the creation and restoration of images much faster.
Suggest creating a specifically named Program Files folder in whichever partition you choose to install your programs and then all you need do is to change the drive letter to that partition letter when the installer gives you the option of where to install its program to. Another little tip is to keep a meaningfully named shortcut to each program in the root of that partition; these can always be easily accessed and copied/pasted elsewhere as needs be - and even after reformatting the OS-partition many of them will still work without reinstalling the individual programs themselves.
Definitely have one or more partitions for your personal data and files and agree with keeping all your downloads and installers on yet another partition (where a similar strategy to keeping shortcuts to all the installer exe files in the root can be helpful later on). The "final" partition would be one to store away your images or ghosts.
Finally, suggest re-lettering your CD/DVD drives to X and Y immediately after installing WinXP and they will then not split up any other hdd drive letters - unless you want them to of course.
Sylvander
05-01-2004, 04:55 AM
PAUL
You said:
The cluster size only really defines the possible size of the total partition for any given file system - the redundancy of lost or wasted space is of much less important with today's enormous storage capacity.
I would say it does more than that.
1. If you sit and watch the defragmentation of a large partition full wave files, where every file is made up of HUGE numbers tiny of clusters:
It is so frustrating to see every file take so much time and effort to defragment, when if the clusters were much bigger and the file was made up of relatively few, the work would go much more rapidly.
2. Then if you use huge clusters for tiny files, the wasted space accounts for a major proportion of the total used space on the partition.
I just dislike waste. It usually leads to inefficiency.
Horses for courses I say.
If you wanted to transport umpteen tanker-loads of milk it doesn't make sense to fill millions of pint cartons to do it.
If you want to deliver to millions of customers each requiring a pint or two, you don't use gallon containers for the job. That would be less than sensible.
If you did, each container would be 70% or 80% empty, just like small files in HUGE clusters.
To repeat, waste never did anyone any good, and it doesn't take very much effort to avoid it. Just the knowledge and will to do it.
Paul Komski
05-01-2004, 07:08 AM
If you use very large clusters then many more individual files on the volume can fit into the sectors of a particular cluster. Result - more wasted space OK - but you will have a greater number of contiguous files (they must be contiguous by definition if they are in one cluster) with less defragmentation and thus better/more efficient file access and much improved data recovery in the event of disasters.
If you use very small clusters you can limit the potential size of the partition, reduce the slack space in each file to a minimum but will have fewer contiguous files and more defragmentation.
Both of the above are particularly true for FAT and are much less important under NTFS, which handles both very small files and very large files very efficiently and has enhanced recoverability. Defragmentation should still be done occasionally under NTFS, but it is much less critical for performance.
I look at this way. If you have a loose-leaf folder with billions (yes billions) of pages in it. There will be some white space at the end of every page that holds a complete topic (or else at the end of each topic or chapter that spans multilple pages). Each page is obviously represented by one cluster. If you make the pages very small in attempting to reduce the total quantity of white space at the end of each topic well, yes, you have saved some storage space - but it will be now much harder to order and to reorganise your loose leaf folder, particularly if the topics in it do become fragmented throughout the folder. Reading the same pages into memory will also be less efficient.
It all really depends on the average file size and the required size of the partition. The default values weren't chosen randomly and without reason. The topic was posted by a WinXP user, and which is why I would particularly recommend using NTFS, when all of this becomes even less important.
If you go back in time to when the total memory available was say just 48K then space is at a complete and utter premium of course. Nowadays there are billions of bytes to play with. I dislike waste as much as the next man - but I like to have things as uncluttered and as efficient as possible as well. We all have different priorities I guess.
PS Final point about defragmentation taking a long time:- The less often you defragment the more fragmentation will occur at an absolute level. Fragmentation helps to cause fragmentation. So the advice is to defragment often and then it wont take that long at all - unless you have very full and very large FAT partitions of course. :D ;)
kmlevine
05-01-2004, 09:03 AM
Thanks to all for the helpful detailed advice. One more question. PArtition Commander doesn't work with a SUB keyboard and moouse setup. I had kind of a freaky scene stuck mid-partition process without ant input control. Any experience with Partition Expert or Partition MAgic regarding USB keyboards?
Thanks again.
Kem
Sylvander
05-01-2004, 12:53 PM
PAUL
I'll need to leave the NTFS aside because I know nothing of its operation.
I believe I've understood what you are saying. Thanks for the explanation.
I'd add this:
I figure there is a kind of ideal number of clusters to a file.
Let's say there were 4 clusters to a file.
Then the maximum wasted space would be [just less than] 1 cluster in 4, or about 25%.
On average they reckon it's half a cluster that's wasted, making 12.5%.
That's too much in my book.
If there were 10 clusters to a file, then the average wasted space would be half of 1 cluster in 10 = 5%.
That's not a bad figure.
50 clusters to a file would give 1% wastage.
Those figures give a rough guide of how many clusters to a file you might settle for.
---------------------------------------------------------------------
As you were saying, the other side of this is that there is a greater tendency for files to become non-contiguous, so there is a need to avoid too many clusters to a file, or be prepared to defragment frequently.
Keeping the waste reasonably low & defragmenting is the way I'd go.
---------------------------------------------------------------------
So if you gain some idea of typical file sizes it's possible to use such knowledge to strike a balance that hits the middle ground.
Whyzman
05-01-2004, 03:12 PM
Perhaps a bit late in the game, but I've found this read to be rather informative:
http://partition.radified.com/
Here's from an earlier discussion with some more insights:
http://www.pcguide.com/vb/showthread.php?s=&threadid=20944
vBulletin v3.6.1, Copyright ©2000-2012, Jelsoft Enterprises Ltd.