A Common True Image Cloning Problem

Been cloning drives for almost 8 years now. When NTFS came along, it introduced a new problem - the metadata files and MFT reserved zone. Properly placed, these elements are optimized by Microsoft algorithm to be located near the center of the hard drive.

A perfectly optimized drive, when cloned by TrueImage (any version!), when asked to clone AS IS to a same size dive, does not do AS IS. To me, "AS IS" means exactly that - all data is to be moved exactly to the same location on the drive that it came from.

It always moves those files smack dab up against all the other files. This then contributes to subsequent fragmentation unless these NTFS files are restored to their proper location.

The only way to do that is with a defrag program that allows off line optimization on a reboot.

Since I, as a cloner and not a backer-upper, am a minority in this group, I was just curious to know if anyone else has paid any attention to this situation. I call it a deficiency, but I can correct it. Maybe it is more of an annoyance.

But my routine after cloning a drive with TI to a like drive (a total duplicate) is to immediately run an offline defrag on a reboot.

 

I'm pretty sure that Ms OS, at least XP, doesn't put the big MFT zone in the middle of the disk but farther forward, based on the notion that most Hdisks are not fully occupied so the physical middle is not the with the lowest average access time to other data locations.

You can run a program like Raxco's Perfect Disk, which forces the MFT Zone to the middle, but I doubt you will ever see any difference in program operating speeds. If you did, in fact, if you disk isn't very nearly full, you would probably see your aeverage access time to data become slightly slower -- but luckily the diff would be imperceptible under most programs.

 

TI does not clone or restore a drive by making a complete duplicate of every sector on the drive. At least this is true for file systems that it understands. TI only copies in-use sectors, and substitutes placeholders for hiberfil.sys and pagefile.sys. Therefore, when a disk is restored (or cloned) it is not an identical copy, and your observations about the positions of files is correct.

I've also noticed that the position of the MFT zone is moved after restoring an image (I don't do cloning like you do). After restoring a partition image that was about 1/4 the size of the partition, the MFT was moved to about 1/4 of the way from the start of the partition. The default location of the MFT with a new XP install is closer to 1/3 from the start of the partition.

I'll second Shieber's comment about the change in performance by moving some of these system files around. On my laptop with a fairly slow 4200 rpm 1.8" HDD, after doing an offline defrag with PerfectDisk and having the MFT and the hibernation file relocated to the middle of the disk, things did get a little slower. The time to save state to the hibernation file increased from 14 sec to 17 sec after the file was relocated. This is a 21% performance degradation.

So having TI pack some of these files closer to the beginning of the disk is maybe not such a bad thing.

 

Those are good points. I happen to use PerfectDisk 8 - the best defragger out there, and have had extensive chats about MFT zone and metadata file locations with their tech guys.

When those files are butted up against the regular files, then that actually creates frag situations where a file has to be split in front and in back of the zone.

Actually, I would like that area to be at the END of the drive. Then it would open up more of the drive to optimized files.

Anyway, I have learned to live with it ever since NTFS came out with XP.

Vista is another matter, especially if there are unmoveable restore files here and there. My fix for that is to turn System Restore off. With one or more duplicate drives always ready to go, that feature becomes not necessary.

 

Depending on how full your disk is, you're going to spanning the MFT Zone, even if it's in the middle.

In certain tests, you can show how the placement inthe middle will tend to be faster. However, the only real world diff it makes is how long it takes to move from the zone to data, so the shorter the average distance between the zone and the data, the faster the average access -- of course that assumes that all used data sectors are accessed equally frequently, which is extraordinarily unlikely. But assuming the data is accessed equally frequently the, fastest location for the zone is in the middle of the data sectors -- that's the shortest average distance. Most folks that I know have maybe half their disk used, in which case, MS's choice of location will work out to be faster than Perfect Disk's choice, which is the reason MS chose to set things up that way.

You are likely to have no more than one file fragmented if it is right up against, and must span, the zone -- that's the physics of it, so it's not like the zone location is going to make lots of files fragmented (or even 2 files) or any one file have more than 2 fragments. And one file being in 2 fragments is not a diff in access time anyone is going to detect. Heck, even a half dozen files in several fragments isn't something that will slow things down enough to notice while you blink. Very simply, what Perfect Disk is pushing about it's program is technically correct if taken within certain strict parameters -- so it isn't snake oil -- however, for most real world intents and purposes, it's just bullsh--. I love Perfect Disk. I think it's a great program but this stuff about zone location -- it's smoke and mirrors in the real world. It's like, should the on/off button be convex, flat or concave. Studies show that concave can be actuated 0.004 ms faster on average than convex. Getting a concave button won't change the world as you know it.

 

Nice explanation, Shieber. As you can notice, I'm also a PD8 fan. Another issue with file placement strategy besides access time is the disk transfer rate, and the area of the disk with the fastest transfer rate is right at the beginning of the disk. PD8's file placement strategy is to put the files needed to boot the machine at the beginning of the drive, followed by seldom-modified files, etc. While this is a good strategy to optimize boot time, it results in a non-optimum hibernation time with the hibernation file placed near the middle of the disk.

On my laptop the main mode of operation is to go into and out of hibernation very frequently. Rebooting occurs very infrequently, maybe once a month. It would be nice to be able to place hiberfil.sys at the beginning of the disk instead of the boot files in order to minimize the time needed to hibernate. Does anybody know if this can be done?

CorkyG, if you're turning off Vista's System Restore then you are giving up one of Vista's best new features - the Volume Shadow Copy service and the implementation of "Previous Versions of Files" that let you go back in time to restore a file as it existed on any previous date (within the range of the Shadow Copy storage). Since this is a forum about backing up, having a month's worth of shadow copies to supplement a True Image backup is a very nice feature. The shadow copies can fill in the gaps between the last disk image and the present date.

 

With a duplicate drive ready to go, there's not much point in System Restore. I used System Restore a lot in XP - and it was never a problem - but Vista's shadow copies started appearing all over the drive and messing things up, so I eliminated them. It's not a problem.

If the shadow copies were not made immovable, then I would probably keep that capability. But, right now, it brings nothing to my table except drive chaos.

 

Hello All,

Thank you for using Acronis Disk Backup software

We'd like to inform you that the ability to create so called "RAW images" was implemented in Acronis True Image 11 Home.This option lets you copy used and unused hard disk sectors. The same sector layout will be recovered if you decide to restore the data from this type of image.

You can download the trial program version to check if this feature is suitable for you.

If you have any further questions concerning our software, please submit a request for technical support or post any of them on this forum. We will do our best to help you as soon as possible.

Thank you

--
Eugene Bogdanov