vi_edit,
<<DIP does not work with a single partion/drive... unless you are backing up to CDR. One 20 gig partition that was half used would still require around 5 CD's to hold the image. Of course, this is all providing that you have a CDR(W).>>
The CDR, or a simlar external medium, is an absolute necessity for any user who cares enough about their data to back it up. Simply sending the data to another partition is a useless gesture in the face of fire, theft, hardware failure, data corruption, accidental deletion, mischief, runamuck programs, etc. Since the partitioning advocates' last remaining "advantage" -- drive imaging -- is so important to them and since they carry it out so often, it stands to reason that they would back up such important information to a safe external medium. If they don't, they're only taking half measures, and in that case, why not just save the Windows registry and prevent 99% of the problems without all the hassle? You're also forgetting DriveImage 4 Pro's Image File Editor feature, which means that an image file need not be 20G or wipe out an entire 20G partition every time it is restored.
<<As far as price goes, I just arranged a group by of Drive Image Pro, which includes Drive Image 4, Partition Magic 6, and boot magic for the price of $30 per person. Expensive? Hardly.>>
More expensive than the $free drag and drop of two .DAT files to achieve basically the same benefits.
Radboy,
<<Very entertaining. I especially like the way you magically turn your baseless opinions into "established facts". Good trick.>>
And I like how you magically turn the words "you", "enough", and "performance" into "u", "enuf", and "perf", respectively. It lends a certain flair, an intellectual distinction worthy of Pee Wee Herman himself.
<<The Windows defragger is a piece of crap. I don't know a single person who uses it.>>
So I prove how an absolutely free tool (MS Disk Defragmenter with Intel Application Launch Acceleration technology) built into Windows accomplishes everything you do with silly redundant partitioning, and your all-encompassing response is to call it a piece of crap. Priceless.
Actually, the only significant disadvantage to Windows Disk Defragmenter is its speed. But since defragmentation is only a twice-a-month walkaway job, that's irrelevant.
<<waste even more $ by burning disks>>
CD
RW.
<<Do you use this method yourself?>>
No, because the entire imaging argument is ridiculous for anyone other than a paid beta tester. Unless you're a software download maniac or an incompetent fool, it's highly doubtful your system is in enough danger of total Windows meltdown to justify the time, effort, and money needed for a weekly drive image; a proper system reinstallation only takes a few hours, and wise people always backup their important data externally for easy retrieval.
If we assume Windows dies on us every six months on average (nevermind that it usually lasts much longer and you'd have to be pretty harsh on your system to hose it that quickly), then the time and effort needed for even a simple registry backup doesn't pay off. Imaging takes even more time. Looking at it objectively, it's a little paranoid.
The most sensible strategy for total system performance and data integrity on single-drive machines is also the least complicated:
1) A big single partition.
2) Defragment twice a week and allow IALA technology to group files in their optimal location on the physical disk surface. Schedule this at night or some other time when it won't bother you (Windows already does by default.)
3) Make regular backups of your
data to some external medium. (Any sane backup plan must include this.)
4) Keep all the original discs for your software in the same safe place as your data backup.
<<Re: "Descriptively named nested folders are an equally effective way of organizing and categorizing data." I disagree, as do many others. This is the Modus methos. If it works for you, fine. For me, partitions are better.>>
No, no. I'm not asking you why orange is your favorite color. I'm asking you why you think a certain technical practice holds a tangible advantage (ie. not just a warm fuzzy geek-macho feeling) over an alternate practice.
A folder named "STUFF" is just as versatile as a partition named "STUFF", and can hold exactly the same. . . stuff. In fact, the folder is superior because it can expand and contract to fit whatever it holds; the partition is rigid. You claim partitions make it easier for you to organize your data. Why? How? It doesn't make sense. Please explain what they can do to organize your data that folders can't.
<<Thx for conceding on the dual/multi-OS point. Are there any other points you've ever conceded?>>
I conceeded on that point before you were an glimmer in this thread's eye -- as opposed to your method of conceeding, which is to simply pretend an entire point never existed and fill the void in the discussion with incoherent rambling.
<<What various "far-flung data" are you talking about?>>
Any data you access that is not stored in your cosy little 2G partition. And don't tell me you never use anything but what you put in your 2G partition, or you would only have a 2G hard drive! Obviously, the other files are on your drive for a reason. You need them there. You use them. And whenever you do, the seek times make your partitioning scheme irrelevant. When you don't, a Windows Defrag is all it would take to keep your frequently accessed main files away from the others, back at the physical beginning beginning of the disk.
<<We have already seen with HD Tach that STRs are max at the beginning of the drive, and that access times are lower if the drive doesn't have to seek full stroke .. to the other iside of the drive. I don't know why you're having such a hard time with this.>>
If you can't wrap your mind around this simple explanation I've given about five times now, there's not much more I can say. I'll write it one more time, but after that, you'll have to ask some one more friendly to phrase it differently or something, because you're obviously suffering from sort sort of bizzare, self-induced mental block:
The fact that data stored at the beginning of the drive has a faster STR is acknowledged but basically irrelevant to real world performance. The claim that a small OS partition will increase system performance in real world usage over a single large partition is incorrect for three reasons: (1) it fails to account for the necessary seeking back and forth between partitions to constantly retrieve needed documents, digital media files, game data, and so forth, preventing drive seeks from taking place in a contained physical area and (2) it refuses to acknowledge that this optimization is already done by any modern disk defragmentation software and (3) it has never been proven by an indepedent benchmark.
Satisfied?
<<I strongly advocate the use of a 2nd drive to receive/store the image>>
Hang on a second. This is breaking news. You're finally acknowledging that it's better to use some kind of other medium to hold the drive image? You do realize that this destroys any need for partitioning, don't you? Because now the only leg you have to stand on is your claim that partitioning increases system performance, which has already been debunked.
LocutusX,
<<a friend of a friend told me that when formatting an NTFS partition, to use the command line switch which uses larger cluster sizes (instead of the small 512 byte), because larger cluster sizes would increase performance when dealing with the large "digital media files" which are so popular nowadays. Is this true?>>
OK, this is a long explanation:
It is given that a smaller cluster size means more clusters per volume and a larger cluster size means less clusters per volume. Larger cluster size increases wasted slack space, but we have seen that today's data is not nearly as susceptible to the phenomenon as before, so cluster sizes could probably grow further, within reason.
Now, on the face of it, smaller cluster sizes don't seem to have any disadvantage. However, in any traditional FAT (File Allocation Table), files are tabled with the number of their first cluster. For example:
[ShaftTheme.MP3] [4,249,906 bytes] [first cluster at #960,822]
When WinAmp loads this song, it asks Windows to get, say, the first 64k of the file to put in a sound buffer. Windows checks the FAT, where it sees that the file begins at cluster #960,822 on this volume. Windows then simply multiplies 960,822 by the byte size of one cluster, say 8096 bytes (8k), comes out with 7,778,814,912, and realizes from that number that ShaftTheme.MP3 is stored roughly seven gigabytes into the volume. It uses the number to tell the drive exactly where to seek and begin a read operation. The drive itself is completely ignorant of clusters. The data is retrieved an passed along to Winamp. Can ya dig it?
Now, you can quickly see that the more clusters we have, the larger and more complicated a FAT gets. And FAT overhead due to size and complexity can slow down any high-level drive access done through the operating system (like, everything.)
In the real world, doubling cluster size (and therefore halving the number of clusters) seems to yield a roughly 10% speed gain on disk benchmarks with smaller gains on real world tests. This is only from some informal benchmarks done by a StorageReview poster in the link provided by Ornery above.
NTFS is a whole new ball game, though. There's a reason Microsoft made 512 bytes the standard cluster size. NTFS is much more advanced altogether. It implements fancy FAT searching methods and better structures that allow it to survive huge volumes with insane numbers of clusters without choking.
Stick with the default 512 byte clusters under NTFS.
<<I suppose that by the end of this gargantuan thread, only the "experts" are left standing>>
Well, before you call us experts, consider this thought: for a couple people who, by the very nature of our argument, must care very much about efficiency and time management, Radboy and I have spent many hours over the span of many days in a futile effort to convince eachother of something which we know full well we can never convince eachother of.
Expert who?
Modus