which is the best zone to put mft+folders?

[Merit Support]

It is related to file data corruption in the outer sectors, not total disk failure.  Microsoft said that before moving the MFT file away from the edge, the disk drive was often blamed as the root cause but the disk drive manufacturers said their drive passes their low-level tests just fine.  At some point, they wondered if the MFT behaved more like a large database so they moved it and the problems went away.

Can you provide link to the original article? I tried searching for it and found nothing ... and regarding what I remember reading myself at MS sources about the topic the reason was speed (but I couldn't find these either).

Sorry, I have no link.  As I explained in an earlier post, I learned this from a Microsoft NTFS guru at a local IT Professionals meeting.  I am in Charlotte, NC where much of the diagnostic and trouble analysis is done.  We are blessed to have such fine speakers at our meetings because these folks are local. When I get a chance, I will try to find my notes from the meeting and give you a name.  (I know what he looks like but that would not help you.)

You may have hit on something there.  The Microsoft speaker works primarily on troubles since NTFS is not really changing now.  It could very well be that the MFT was relocated for performance reasons and the unintended consequence was that MFT corruption was nearly eliminated.  He said that he had previously spent much time repairing MFT files but now that the MFT has moved away from the outer edge the problem is very rare.

I have had similar experience in computer repair.  Replace keyboard for keyboard error. Replace mouse for mouse freezing.  Finally replace PSU when system fail to start.  After replacing PSU, reconnect old keyboard and mouse.  System works fine.  Original PSU never did test bad.

[Leolo]

Well, confirmed.

I have done a few installations of several versions of Windows under a virtual machine with a 80 GiB hard drive and here are the results:

- Windows 2000 SP4 always puts the MFT at the beginning of the disk (at the 4th cluster, which means 16384 bytes with default cluster size)

- Windows XP SP3 always puts the MFT at cluster number 786432 (using a default 4 KiB cluster size this means exactly 3 GiB)

- Windows 7 SP1 always puts the MFT at cluster number 786432 (using a default 4 KiB cluster size this means exactly 3 GiB)

I repeated the tests with a 750 GiB hard drive, and I got the same results.

No matter how big the hard drive is, Windows will never put the MFT beyond the 3 GiB mark.

Does anyone know why??

[poutnik]

I know I have read several times at various sources - links not handy,
that keeping MFT with 3 GB offset brings about 10% performance gain
and was said to be Microsoft recommendation for performance reasons.
At some place it was lower value of either 30% either 3 GB.

[Leolo]

Poutnik, you probably read it here:

http://wpblog.perfectdisk.com/?p=350

But Microsoft has since removed their article and it's nowhere to be found in their website. It's a pity

[rdsu]

The best place to put the MFT and Directories is in the middle of your most used data...

[jeroen]

It sounds plausible to me that reliability is improved if the MFT is placed a bit into the disk, instead of at the beginning. I will therefore change the datadisk scripts to place the MFT at 3GB. Not at 30% into the data, like on system disks, because for datadisks it is fastest to place the MFT at the beginning of the disk (for several reasons). It will only improve reliability on data disks that are physically at the beginning of a disk, perhaps I can make it conditional on that. There will be a marginal speed penalty, but that is a small price to pay for better reliability.

[rdsu]

It sounds plausible to me that reliability is improved if the MFT is placed a bit into the disk, instead of at the beginning. I will therefore change the datadisk scripts to place the MFT at 3GB. Not at 30% into the data, like on system disks, because for datadisks it is fastest to place the MFT at the beginning of the disk (for several reasons). It will only improve reliability on data disks that are physically at the beginning of a disk, perhaps I can make it conditional on that. There will be a marginal speed penalty, but that is a small price to pay for better reliability.

I don't know if will change the data disk script because what you read at PerfectDisk blog, but PerfectDisk place the MFT at the beginning of disk for data disks, or beginning of partition for data partitions...

[Leolo]

rdsu, are you sure about that?? What you are saying contradicts what the official blog of Raxco says.

Do you really mean that the information in the Raxco blog is incorrect??

Remember that the first fragment of the MFT (about 16 kb) is unmovable. Did you take that into account when measuring the MFT position in the disk?

Regards.

[rdsu]

rdsu, are you sure about that?? What you are saying contradicts what the official blog of Raxco says.

Do you really mean that the information in the Raxco blog is incorrect??

Remember that the first fragment of the MFT (about 16 kb) is unmovable. Did you take that into account when measuring the MFT position in the disk?

Regards.

What I said about PerfectDisk if for data disks/partitions, not system disk/partition...