Whirlpool

Hey

I've read somewhere that you can create something called a PAR (of about 10%) when burning data discs (dvd) that means that even if you lose 10% of the data (from scratches on disc or something), the full disc is still recoverable.

Is this true ?

Can someone please show me how it's done.
I just wanna back up some important data (mainly videos) onto dvd's, but wanna make sure that they're pretty safe (even if they get slightly damaged after some time).

Is this PAR, an archive file extension or does it stand for something else ?

Thanks.

I got WinRAR and now QuickPAR also.

D1splayNam3 writes...

Can someone please show me how it's done.

WinRAR has its own system, recovery record - when creating an archive, check the box in the General tab, and the percentage in the Advanced tab.

In quickpar, just select the files and percentage, and away it goes. It produces a bunch of .PAR2 files to save with your data.

I'd be inclined to think the pars would be slightly more robust a system, although I don't use either for my own uses.

Thanks

So with QuickPAR, I just make all those .PAR2 files and just burn them with my data disc ?

Should they be in a certain folder or anything ?

And, if my data does become corrupted, how do I use the PAR files to recover it ?

D1splayNam3 writes...

Should they be in a certain folder or anything ?

They don't have to be.

And, if my data does become corrupted, how do I use the PAR files to recover it ?

Quickpar does that. It will scan the files for errors and repair from the par2 files where necessary.

Ok. Thanks

Would you recommend WinRAR's system or QuickPAR ?

Also, do the settings I've used look ok ? I got nfi what half the stuff mean.
img172.imageshack.us/img...15/sshot2qe6.png
img413.imageshack.us/img...18/sshot3tg3.png

D1splayNam3 writes...

Would you recommend WinRAR's system or QuickPAR ?

As I don't use Winrar myself I don't know if there is anything proprietary about it's method. Also Quickpar works on all file types, compressed or otherwise, so I'd say Quickpar.

Another thing that comes to mind, in my experience when a DVD is cactus, a file will copy across until it gets to the bad spot - and nothing else can be salvaged after that, even with dedicated recovery programs. Under this scenario, if one of your 1 GB files carks it at 200MB, you won't have enough pars to fix it.

So as just an little extra step, you might like to make multi-volume RARs (say 50-100MB or so, using no compression), and then produce the PAR set from those files. That way if it's just one file that's a dud, you should have no problems restoring it.

I'm working with smaller files - mostly videos from about 100MB up to 700MB.

If one of those files are screwed, won't the PAR set save me ?
Say I'm using 10% redundancy

Your better off burning 2 copies of the same backup data. Its a lottery that the par file will recvoer the file you may lose.

D1splayNam3 writes...

If one of those files are screwed, won't the PAR set save me ?

It depends solely on how much of the file you can salvage, which is infinitely variable depending on what has caused the corruption.

My suggestion was to break up the file into smaller pieces, which may (or may not) limit the damage done by a faulty/scratched disc.

Parity volumes (.PAR) are more suited to recovery of data that gets corrupted on the fly, like on the Internet, or like Nasa uses for its Deep space probes, where along with the data (which can be VERY FAINT in signal strength) they also send Recovery data for the transmission, and in MOST instances they can recover the data thats lost due to signal strength or interferance.

Its COMMONLY used on Usenet for when numbers of blocks are unavailable for some reason, you can still recover the lost blocks, assuming you have enough Parity volumes.

The software technology is based on a 'Reed-Solomon Code' implementation that allows for recovery of any 'X' real data-blocks for 'X' parity data-blocks present. (Data-blocks referring to files OR much smaller virtual slices of files).

For those that care.

en.wikipedia.org/wiki/Reed-Solomon

Thanks all.

I've attempted to create a backup par set with QuickPar.

My data was a couple 700MB videos and some rarred (split-archives about 15MB each file) files also.

I tried to copy the entire folder into QuickPAR, but it looks like it only accepts files. So I had to copy every file over separately. I left all the other settings to the default and it then created the par set. All seems to have completed well.

Anyways, if my data does get corrupted or the disc slightly scratched, how will I know ? And how do I check it ?

To check it, just click on one of the PAR2 files to open it. QuickPar will read through the original files to verify that they are still intact. If all is well, the message in the progress bar will say "Repair not needed". If something has become corrupt and you have enough parity data to repair the damage, it will give you an option to rebuild the files.

PARS are a waste of time and space

The only reason PAR files are used on Newsgroups
(thats the only place that seems to use them ) is that the RAR files get removed

If people actually STOPED using .PAR files their would be more room for the .RAR files

Burn the data on decent DVD media - TY/TDK
You could always burn it on multiple cds or dvds so you have a multiple
copy if its that important

If you have the data as .RAR files along with the SFV then thats all you really need

Burn at a SLOWER speed for more reliable burn (2x 4x)

PAR info: en.wikipedia.org/wiki/Parchive

Thanks for the info

tafed00d writes...

PARS are a waste of time and space


How are they a waste ?

If I understand correctly, if you create par files with 5% redundancy for data that you've burnt to a dvd disc, then up to 5% (any 5% that is) of the disc can get damaged (ie. scratched or whatever that would give a CRC error when trying to recover it) and you will still be able to recover it.

Is this not correct ?
If it's correct, then they are definetly NOT A WASTE OF SPACE OR TIME. They're fantastic.

But I'm not sure if I've understood correctly. Sounds too good to be true.

Also, what happens if the disc gets scratched on the parts where the par files are and they aren't complete. Are you kinda screwed then if you need to recover the actual data you want ?

The PAR file set doesn't have to be complete. They can sustain a bit of damage and still be of use. Some can be missing altogether. However many blocks of parity data remain undamaged, that's how many blocks of the original file set you can recover.

That's why they are so good for internet data. After downloading a 1GB RAR file and discovering that there was a glitch somewhere in the transfer which makes the whole thing useless (and no, not always because some of it was deleted from the server), QuickPar can tell you exactly how many parity blocks are needed (usually only one or two) and you only need to download that number of blocks from the PAR set in order to repair it. The first time a PAR set saves you from wasting a gigabyte of download quota, you'll see why they aren't a waste of time and space.

Hi D1splayNam3, here's an example of how to use PAR to recover from faulty media:

1. You have 9 DVD disks filled with data, each disk contains some files, all in the root folder (eg 7x600MB files). Currently all the disks are OK, but you want to be in a position to be able to loose any 2 WHOLE DISKS at some point in the future and still be able to recover everything. You will need two blank DVD disks.

2. Copy all the files from each disk into a single folder on your PC. This will take about 40GB or so, and you will have all 63 files in a single folder.

3. Use QuickPAR to create the PAR2 files based on the following criteria:
Total source files size: 40GB
Total Recovery Data Size: 8GB (e.g. 2xDVDs),
Number of PAR files: 10 PAR files (put 5 on each disk)
Recovery Block Size: 30MB
This will take about 14 hours to compute (on my P4 2.5Ghz)

4. You will be able to loose about 8GB worth of data and still recover it with what wasn't lost.

Thanks for that. Very well explained.

But I'm not dealing with things that big. The most I deal with is about 4GB at a time (as I back up to standard DVD-5 discs). So I think I'll be filling the dvd with about 4 GB data and then creating par set at roughly just under 10% (so that it fits on dvd). That should be alright yes ?

Also, you know how you said, 10 PAR files (put 5 on each disk) , if I only lose a little bit from one of the discs, will I still need the other par's from the other discs to recover it, or just the par's from that disc ?. . . or does that depend on exactly how much I've lost ?

tafed00d writes...

If you have the data as .RAR files along with the SFV then thats all you really need

Aren't sfv files there to just check that files are good and complete? They don't repare anything iirc. .. so they'd be useful for backing up downloaded files, because I could re-download the corrupt/missing parts at a later time if they get damaged

However, if the files are my own and they become damaged I don't think an sfv would help me. QuickSFV would just tell me 'your files are incomplete', but I'd have nowhere to get the missing pieces, right ?

Burn at a SLOWER speed for more reliable burn (2x 4x)

Yeah. I'll do that with important stuff. I usually go with 4x anyway, unless I'm in a hurry.

D1splayNam3 writes...

So I think I'll be filling the dvd with about 4 GB data and then creating par set at roughly just under 10% (so that it fits on dvd)

this all depends on what you are trying to protect against, and after a failure, whether you lose the whole disk, or just 1GB worth of data from it. So if you have 1x DVD at 4GB and make a 10% PAR2 recovery sets and then put them on the same disk, you will be able to lose about 10% of the data and recover it, provided you still have 90% of the remaining data and the PAR2 files left. Since the technology is scalable, I suggested making a PAR2 recovery set across all 10 disks, meaning the recover data will fill a whole disk. (allowing one disk to fail). I specked my example above allowing you to lose 2 disks completely.

My experience has been when you lose a DVD disk, the whole disk is dead after that, hence my advice to design your data recovery via PAR2 technology to use this as the model.

Ok

MRIS writes...

I suggested making a PAR2 recovery set across all 10 disks, meaning the recover data will fill a whole disk. (allowing one disk to fail). I specked my example above allowing you to lose 2 disks completely.

If I do that (Backup 10 disks with 10% redundancy and put 1/10th of the par2 files on each disc), then if I do lose 2 discs, I will have lost 20% of my files aswell as some of my PAR2 files. . . so I don't think I'll be able to recover the files off the 2 lost discs seeing as I've only got a 10% redundancy ? Isn't that right ?

Anyways, say I lost 1 disc only, and I've still got the other 9 all good, will I still be able to recover all the data on that other disc, even though I don't have all the PAR2's (because I lost some with the disc) ?

the best way to learn about quickpar and par2 recovery volumes is to actually have a play with some test data. Copy a couple of GBs worth of files into a folder, make the PAR2 volumes based on those files using quickpar, then see how by deleting some of the files, the quickpar system can get them back. Learn the % involved and then you will get a good feel on how much recovery data to make, what blocksize to use, etc.
My advice was to keep the PAR2 files all by themselves on their own disk, not distributed amongst the disks they are supposed to protect. Thus if you lose one of the data disks, quickpar gets it back. If you lose the PAR2 disk, quickpar can regenerate a new set of PAR2 files for the replacement disk.
In my example, I said 9 data disks and 2 par2 disks, allowing any 2 disks to be recovered if lost or destroyed.