Windows 10: Should I transfer my Upgrade or Clean Install?

The best way to prevent transfer errors is to use validated transfers whenever you copy files from one storage media to another.

Windows performs some basic checks when files are copied from one place to another, but it does not do a thorough validation. The only way to be absolutely sure that everything was transferred properly is to perform a bit-for-bit comparison between the original and the copy. I don't know for Macrium Reflect, but many backup software packages at least do checksum compares.

 

rest assured, Macrium will not let you down. Once you get the backup it works. Do not lose sleep over it.

 

Lol I already backed it up! I just need to ensure all my stuff is on both drives before I wipe the 2tb.

 

Do you know what kind of validation Macrium Reflect is doing - if any?

 

Yep, only checksum compares for Macrium, not a full validation - http://knowledgebase.macrium.com/dis...res?src=search

 

The help page only says that auto-verify only is available with the full edition of Macrium Reflect. Does the free version of Macrium Reflect offer file verification at all?

Second point: If Macrium Reflect succesfully verified an image file, there is a very high probability that the image file is in good shape. There however still is a (very small, if you ask me) probability that the file verified, although it in fact is not a copy, which is 100% identical. Macrium Reflect cannot prevent that,

 

Yes: Macrium Software | Macrium Reflect Free

That's precisiely what I meant by no full validation. The further step of confidence I have gotten, when needed for critical imaging, is by actually trying restores on a given system AFTER I've taking a redundant image with other software in which I have confidence, e.g., Image for Windows/DOS/Linux. I would go so far as to term those instances "spot validation" or "validation sampling".

After all, one full validation (often 3 batches) is all you need in the pharmaceutical industry I work in, at least until you change the process and/or it's equipment.*Wink

P.S. no, I don't test critical restores 3 times because the U.S. FDA doesn't require me to.*chuckle

 

That page now does not only state that verfication would be included, but also "auto verification" would be in. That contradicts what their other website claimed.

Yeah, I guess it's pretty safe to consider a file to be a true copy, if it verified. Chances it is not should be very, very small.

 

We are talking about two different procedures with Macrium. Clone and Image.
The OP used the Clone procedure which copies everything from one disc to another. I do not know the procedure for verifying the Clone on Macrium site as I have never used the Clone procedure. The Image procedure makes a compressed copy of the disc or partition and saves it as a file. I have used an Image to transfer my system to a new drive. Macrium uses a hash method for the blocks it copies and has a option to verify the block hashes on the saved Image. In v6 of the free version under Advanced Settings you can select to auto-verify the Image or you can run a manual verify when ever you chose.
I would suggest to the OP to make an Image with Auto-Verify of his old disk and save it if he ever needs it. He should also make an Image of his new disk.

Jim *Cool

 

Yeah, MD5 is a 128 bit hash so the chance of 2 files having the same MD5 is only 2 ^ -256.

That is a small number. Possible but on average only 0.000000000000000000000000000000000000000000000000000000000000000000000000000000000086% of the time.

That is 76 zeros (hopefully).

 

Cool, thanks for the clarification!

 

Pretty much the same concerning verification - http://knowledgebase.macrium.com/dis...Cloning+a+disk

 

That is what I call very, very unlikely.

Another aspect is the question, which kind of corruption leads to which change in the hash. Harddisks only store ones and zeros. Depending on where inside the file an error happens (a bit is changed), this might have a different effect on the hash. It would then be interesting to know, what exactly would have to be changed in order to create a different file with exactly the same hash.

MD5 hashes have in the past been used for security ("securely" storing passwords for example). I know there have been scientistists, who have written essays on the probability of being able to create a hash collision:
Different data, which creates the same hash as the original.
Usually it does not need one small change in the data, but it needs more or less completely different data in order to get the same hash again.
The probability that such a huge number of incorrect bits are transferred (with the errors at exactly the right places) is another limiting factor.

 

Hoping not to get unacceptably far from the thread topic - I'll just chip in with I think there's a bit more to fully validating image restoration than having a vanishingly small probability that the data was encoded into the image with 100% accuracy. It needs to make a complete round trip across every link in the chain until you're exactly back where you started from. You can only go so far into this before Heisenberg's Uncertainty Principle kicks in.

Granted, mere "verification" serves me fine from a practical standpoint.

 

I have read that link but it only mentions doing a Verify File System on the source disk before it starts the clone. No verification on the target disk after the clone.

Jim *Cool