Par? Par2? DVD? JPG? WTF?
August 29, 2007 4:18 PM Subscribe
I've recently decided to backup all of my photographs onto DVD. But I'm concerned about stuff like data recovery, scratches, etc. So I figured I could add some .par files to the DVD, to be on the safe side. Trouble is, I've never used .par files before.
I've had a quick poke around the web, and discovered that .par2 files are better than .par files, but I'm still a little unsure about the whole thing.
How do .par files work? What's the best method to create them? How many do I need? Is there a guide somewhere that explains how to create/burn to DVD with .jpgs/recover data with them?
I basically want to make sure I don't go to a lot of effort to create something that's not going to work when I need it to. :)
I've had a quick poke around the web, and discovered that .par2 files are better than .par files, but I'm still a little unsure about the whole thing.
How do .par files work? What's the best method to create them? How many do I need? Is there a guide somewhere that explains how to create/burn to DVD with .jpgs/recover data with them?
I basically want to make sure I don't go to a lot of effort to create something that's not going to work when I need it to. :)
Unless you've got a metric arseload of pics, I'd just burn multiple copies of everything. This will also allow you to store them in separate places.
posted by pompomtom at 4:39 PM on August 29, 2007
posted by pompomtom at 4:39 PM on August 29, 2007
Best answer: DVDs are much better than CDs as far as error recovery goes.
Parity pretty much works like this: Let's pretend you have six pieces of data and one piece of parity data (it doesn't matter how big the pieces are - one data block, one DVD, whatever; for simplicity's sake let's pretend the data is only a single bit). It also doesn't matter how many data pieces you have for one parity piece, although since parity can only recover one piece of data, the risk of having more than one bad data piece increases.
Sample data: 0 1 1 0 1 0
To get the parity data, add the bits together, and you get 1.
Now let's say one of the data pieces gets corrupted: 0 1 ? 0 1 0
Add these together (assuming the damaged bit is 0), and you get 0. This doesn't match the parity bit, so the missing data bit is obviously 1 instead.
This is a pretty basic digital-communications concept that's been used for decades in networking, modems, etc.
I Am Not A Computer Scientist, so I could be wrong about this, but that's my understanding of how it works. PAR2 uses the same concept, but splits the parity data into multiple redundant blocks, and the amount of data you can recover is directly related to the amount of PAR2 data you generate (e.g., if you have 100 blocks of data and 10 blocks are corrupted, you'll need 10 blocks worth of PAR2 data).
posted by neckro23 at 5:08 PM on August 29, 2007
Parity pretty much works like this: Let's pretend you have six pieces of data and one piece of parity data (it doesn't matter how big the pieces are - one data block, one DVD, whatever; for simplicity's sake let's pretend the data is only a single bit). It also doesn't matter how many data pieces you have for one parity piece, although since parity can only recover one piece of data, the risk of having more than one bad data piece increases.
Sample data: 0 1 1 0 1 0
To get the parity data, add the bits together, and you get 1.
Now let's say one of the data pieces gets corrupted: 0 1 ? 0 1 0
Add these together (assuming the damaged bit is 0), and you get 0. This doesn't match the parity bit, so the missing data bit is obviously 1 instead.
This is a pretty basic digital-communications concept that's been used for decades in networking, modems, etc.
I Am Not A Computer Scientist, so I could be wrong about this, but that's my understanding of how it works. PAR2 uses the same concept, but splits the parity data into multiple redundant blocks, and the amount of data you can recover is directly related to the amount of PAR2 data you generate (e.g., if you have 100 blocks of data and 10 blocks are corrupted, you'll need 10 blocks worth of PAR2 data).
posted by neckro23 at 5:08 PM on August 29, 2007
Check out DVDisaster. I've never used it successfully (actually I'm working on a test burn that I did on a known-bad disc, and I think it should be able to recover my data, but it hasn't yet.) but it's very promising and worth looking into.
posted by Myself at 8:11 PM on August 29, 2007
posted by Myself at 8:11 PM on August 29, 2007
@neckro23
i was always under the impression that dvdrs were even more vulnerable to scratches than cdrs. the logic being that if both types of media had the exact same scratch the one on the dvd would impacted approximately 14 times more information.
i agree with mrbill, why not make multiple copies of the dvds? you could even keep one off site.
posted by phil at 8:54 PM on August 29, 2007
i was always under the impression that dvdrs were even more vulnerable to scratches than cdrs. the logic being that if both types of media had the exact same scratch the one on the dvd would impacted approximately 14 times more information.
i agree with mrbill, why not make multiple copies of the dvds? you could even keep one off site.
posted by phil at 8:54 PM on August 29, 2007
If you can fit everything on a single disc, I'd simply make multiple copies (but throw some pars on there too-- why not?). If you're using several discs, make entire par discs-- That way, even if you lose/destroy a disc, you won't lose anything. If it's a four disc set with four par discs, you have to lose five discs before you lose anything. If you make two copies of the set, losing the wrong two will do it.
posted by alexei at 1:50 AM on August 30, 2007
posted by alexei at 1:50 AM on August 30, 2007
Best answer: Please don't go with multiple copies in preference to parity files.
If you make two copies, and they're both unreadable in the same place, you've lost that file. With a 100% PAR set, you'd have to lose an entire disc before you couldn't reconstruct your files.
posted by Leon at 3:01 AM on August 30, 2007
If you make two copies, and they're both unreadable in the same place, you've lost that file. With a 100% PAR set, you'd have to lose an entire disc before you couldn't reconstruct your files.
posted by Leon at 3:01 AM on August 30, 2007
Best answer: My quick recommendation.. Take ~2GB of data, create 200MB of par2 data with very small block sizes. Then, when you burn the DVD, put a second copy of the data (and as many copies of the par2 data as you can) in a subdirectory. This way you don't have to wait for huge amounts of par2 data to be generated (very CPU intensive when using small block sizes), but you get really good redundancy. The sub directory thing also helps to put par2 files near the beginning, middle, and end of the DVD - if the par2s are together on the disc, you might loose them together, which would be bad.
Here is a very good previous thread on the topic:
posted by Chuckles at 9:58 PM on August 30, 2007
Here is a very good previous thread on the topic:
Is recorded media such as CD-R completely permanent?(especially my answer :P)
October 14, 2005
posted by Chuckles at 9:58 PM on August 30, 2007
This thread is closed to new comments.
If you need more explanation, I can post with better instructions.
posted by jammnrose at 4:25 PM on August 29, 2007