Slashdot Mirror


Cringely's P2P Backup Idea

gewg_ writes "If Napster and Bit Torrent had a baby, would it Baxter? As a follow-on to Cringely's last column where he talked about having a backup strategy in the wake of Hurricane Frances, this week he proposes a distributed RAID notion as a solution."

24 of 205 comments (clear)

  1. over and over again ... by duplicatedAccount · · Score: 3, Interesting

    Well, we leave the data where it belongs: in the proxy network where the processes live too. Still a bit incomplete, but maturing WebDAV and mountable slices forthcoming...

  2. Freenet by John_Allen_Mohammed · · Score: 5, Interesting

    Just insert a bunch of data into the network.. record the keys and retrieve once a week then delete. That should keep the data retrievable from the network for a good while. Using two nodes would help. Plus everything is encrypted with some heavy shit.

    Or, just make a local-freenet on the company lan.. everything is encrypted and unretrievable without the proper keys, so it's very secure and it's distributed.. + FEC encoding.

    That assumes freenet works, AFAIK it's still fucking broken. Ian Clarke is playing too much politics with the project and the only coder that really understands freenet (Mathew Toseland) is swamped with ideas, day after day.. it just gets worse and worse... The donations seemed like a good idea, but after watching the DEV list for the last 18 months, I realize it's a failed project :(

    --

    Skype Me! username: john_allen_mohammed
    1. Re:Freenet by Joe+Tie. · · Score: 2, Interesting

      Ah, so that explains it. I finally got enough ram to keep freenet going 24/7, and was surprised to find it so unreliable. I wasn't expecting a speed demon, but I was expecting that links to files on freesites would work if the site itself was. That, so far, has seldom been the case. Are there any other similar projects going on?

      --
      Everything will be taken away from you.
    2. Re:Freenet by MrJay · · Score: 2, Interesting

      That assumes freenet works, AFAIK it's still fucking broken. Ian Clarke is playing too much politics with the project and the only coder that really understands freenet (Mathew Toseland) is swamped with ideas, day after day.. it just gets worse and worse... The donations seemed like a good idea, but after watching the DEV list for the last 18 months, I realize it's a failed project :(

      Check out other development lists on popular projects (if they're public). You'll find that heated debates, arguments, and blatant personal attacks are very common. The Linux kernel list has years of flames between Linus, Alan and other major contributors.. but development continues, doesn't it?

      Same in this case. There are two sets of debates on the Freenet lists. The first set includes those with a background in computer science, an understanding of programming, or an understanding of the theoretical concepts behind decentralized communications. The second (and much larger) set of people are primarily concerned with advancing their own positions within the project. Their ideas are based on false assumptions or a blatant ignorance of Freenet's goals. They will propose things that are completely opposite the goal of a "decentralized anonymous attack-resilient network" and then blame Ian's ego when their ideas are shot down (perhaps unapologetically by Ian himself, sometimes by others). No amount of hand-holding or compassion can convince this group that their ideas are simply incompatible.

      All this aside, the stable releases have been working excellent for me and I am running a stock Freenet node; no tweaks. The introduction of NGR also came with lots of bugs, which (imo) is the main cause for the lackluster performance of Freenet. Now that it's working better, perhaps we can judge NGR on it's abilities (for better or worse) instead of the bugs which cause any analysis of Freenet itself to be inaccurate.

    3. Re:Freenet by MrJay · · Score: 2, Interesting

      Entropy is dead and has been for a few months at this point. They decided that Freenet is broken and claimed they could do better. Time did tell.

      The other point many people critical of Freenet make is that other P2P systems are much faster. One chap came into the Freenet channel and claimed Entropy is much better than Freenet because it's written in C++ and real fast. He didn't realize that Entropy development was dead and that the network consisted of about 20 peers. Nobody knows the precise number of Freenet nodes in the stable network, and any estimate I give is just a guess.

      Ever since March 31st/April 1st of 2003, it's just been downhill.

      Ah, The introduction of NGR. The idea is that classical routing was not scalable, so something else was needed if Freenet was to support more users. It didn't make sense when a splinter group broke off from the main project and used the infamous "build 692" to claim that new Freenet development was broken, since the old build works for the 50 or so people who were using it. We are now close to the point where a proper simulation of NGR and Freenet will tell us much about NGR itself, since I believe most of the terrible bugs have been fixed.

    4. Re:Freenet by MrJay · · Score: 2, Interesting

      I think you meant 'unfair' instead of 'inaccurate'. And that still sounds weird. Imagine a benchmark test that came out, showing Intel's newest processor to suck due to a bug. Would you buy Intel saying "yeah, but if you ignore the bug, it rocks"?

      Apples and Oranges. And I meant "inaccurate". Fairness isn't an issue. Your analogy is flawed for the following reasons:

      * Intel makes chips that are fundamentally still based on the x86 model, which has been around for over 20 years.
      * Freenet is around 5 years old, and still not at version 1.0 yet.
      * Intel doesn't benchmark chips that don't work; only chips that are working.

      You've brushed aside the fact that developing an anonymous, decentralized network that is resilient to attack is hard, and much of the theory is untested in the real world, even though mathematically the idea seems perfect. Since you pay hundreds of dollars for an Intel chip, and Freenet is itself Free, what do you expect given that Freenet isn't actually complete? A perfect product? Give us an R&D budget equal to Intel and Freenet will work much faster than the current pace.

      And where did I say benchmark? I said analysis, and what we wish to analyze is how Freenet is currently routing, for better and/or for worse. Once this is known decisions can be made to move Freenet towards a real release. What part of this is unhealthy? And can you even name one other P2P network that scales better than Freenet?

  3. Re:What an awesome idea by Mod+Me+God+Too · · Score: 5, Interesting

    I once encoded some data in a few MP3s... this was back in 2000. The MP3s were long speech files... about 30mb/file @ 160kbps and were popular, but took so long to transfer, so to propegate the 'new' files as quickly as possible I reduced the bit rate from 160kbps to 32kbps and added in the 'extra' 'noise' as I did this - as it's speech it didn't really matter.

    If I do a search now they're easy to find, much easier than the original 160kbps were.

    This was just a test, no special data used - but an amazing way to archive and distribute data.

    --
    --

    It is not the commies, the government, the nigger, nor the corporates. It is your paranoia.
  4. makes my heart a little warmer... by sgant · · Score: 2, Interesting

    From the article:

    It sounds from every description like the solution is Linux-specific, but I'm sure it can be made to work with other UNIX variants, especially since Gmail, itself, runs on Apple xServe 1u boxes. Windows compatibility is unknown, but I'm sure someone will solve that soon.

    I know, it's a little childish, but I get a good feeling when I see something small...even this little thing here...that thinks of other OS's first and Windows compatibility will be "real soon now" or something like that.

    --

    "Leo Fender was in a 'state of grace' when he designed the Stratocaster." -- Paul Reed Smith
  5. I lived the storm by jsm008us · · Score: 2, Interesting

    Well, I lived through this storm, checking my PC upstairs to make sure nothing was going to damage it. If the storm was risking the roof flying off and my room becoming flooded, I would have taken out my hdd. This sounds like a brilliant idea.

    Hey, it beats trying to store data to gmail accounts! ;)

    --

    mysql>SELECT * FROM users WHERE clue > 0
    0 Rows Returned
  6. Save Betamax by chatooya · · Score: 4, Interesting

    Ideas like Cringely's will be impossible if the INDUCE Act passes.

    Save Betamax is a national Congress call-in day this tuesday to oppose the INDUCE Act. It might be our last chance to stop this bill.

  7. Re:damn.. by dotwaffle · · Score: 5, Interesting

    I had this idea in about '97 or '98. I looked around to see if anyone else had done anything like this (remember, this is kinda pre-mass-P2P) and found that someone had done so, but on a business scale solution. I think it was called Mango, and is still in production today. It essentially made a portion of your drive available for a drive letter, then whetever was copied onto it could be seen by all. The data was stored in at least 2 places, so if one went down, there was still one copy, and the remaining copy would duplicate, so that there was always at least 2 copies. In the end, I think nobody went for it because it was too expensive... But this is EXACTLY what a lot of Small-Medium businesses need atm. Bring on the Mango's!

  8. Nice idea, but by moonbender · · Score: 4, Interesting

    It's a neat idea. In a nutshell, he suggests a Peer to Peer encrypted storage network. You get exactly as much storage room as you are willing to offer yourself for others to use. When you store anything, it's encrypted and automatically spread to other systems.

    It doesn't make for a very safe backup, though: What happens if somebody decides to stop the service and just deletes his local storage? You've got no more backup at least for a while, and you might not even know it. And of course, other people have head crashes, too, which would also obliberate your backup at least for the time it takes to recreate it from your own data. Of course, by that time, you might have deleted it yourself, either by accident or knowingly, since you have a backup after all. A viable solution would be to store every file multiple times on different remote servers, although that'd lower the storage capacity you get. It's still the right step, though.

    The crucial problem is that the service provider can't really give any guarantees that you will be able to regain your lost data. With three or more independent copies in different locations, it's very unlikely that the backup won't work for some reason, but a backup that's not 100% is not a very useful one, especially in those situations where backups are really crucial.

    It's still a neat idea, and to my knowledge has not been done to that degree of sophistication. Of course, as others suggest, nobody is stopping you from inserting encrypted data into Freenet, but that's nowhere near as fast and secure as this could be. And while it's not a true backup, it's better than no backup at all, and most likely enough security for many persons.

    --
    Switch back to Slashdot's D1 system.
  9. I still say Gmail... by plasticmillion · · Score: 4, Interesting
    Not to beat a dead horse, but Cringely seems like he was in a bit of a hurry to reject the Gmail solution. Wouldn't simple encryption solve the privacy problem? The Gmail text analysis is based on the assumption that the data is some kind of natural language text, so it would be baffled by anything else. Huffman encoding (or some other compression) would do the trick and save space besides.

  10. If Diablo 1 was in P2P by CrazyJim1 · · Score: 4, Interesting

    If your character data was stored on everyone else's computer, it would act like a virtual server, where if a few data sets get hacked, they'd be corrected by the whole.

    P2P can work in wild ways we haven't even tapped.

    too bad orrin hatch is trying to outlaw p2p:
    www.geocities.com/James_Sager_PA

  11. Re:No thanks by DarkHelmet · · Score: 2, Interesting
    Encryption? Simply using GnuPG or any of the free AES encryptors out there will make it incredibly secure. If your data is sensitive enough, you should be doing this already...

    Or for that matter, why not build encryption into the system itself, so that you don't have to manually do it.

    --
    /^[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,4}$/i
  12. Already a commercial product by YetAnotherName · · Score: 3, Interesting

    A company called 312, Inc. already has a commercial product for P2P backups called Lean On Me.

    I don't work for them, etc.

  13. Cringley's been reading my posts! by Anonymous+Cowdog · · Score: 2, Interesting

    I've also been suggesting this for years. I'm too lazy to search for the older posts, but here is one from July:

    http://slashdot.org/comments.pl?sid=115027&cid=974 3518

    Of course what matters, though, is not talking about ideas, but *doing* them.

  14. Pricing by duvel · · Score: 2, Interesting
    Cringely writes: Apple, for example, will let you mount up to a 100 megabyte iDrive as part of its .mac Internet service, but that costs $99 per year. Eight dollars per month for 100 megabytes of storage is too darned much.

    The company I work for (banking) sells storage for 120 euro per gigabyte per year to our internal clients. That's storage on RAID-disks (think StorageTek and the like), including backup (on tape) and all necessary services (people doing maintenance, restoring backups, etc). 120 euro / gigabyte / year comes to 1,22 dollar / month / 100 megabytes (compare to 8 $ per month with Apple). Considering our 1,22 $ plus some network costs, plus maintaining a billing system for a couple of million clients, and a bit of profit margin, maybe 8 $ per month is not a rip-off.

    --

    I have a photographic memory for numbers. I know almost a hundred of them.

  15. Baxter for Video Activists by lo_fye · · Score: 2, Interesting

    I was thinking about something like this for video activists who frequently have their tapes/discs confiscated by the cops. It'd be great if they had PocketPCs with webcams that were operating in a baxterian sort of way such that the video they were taking was simultaneously being recorded to the storage of other activists/media within wifi range. You could have wifi NAS (network storage) in vehicles and apartments surrounding the demonstration area, as well as on ipod-level storage in future wifi enabled pocketpcs. 3G cameraphones with hard drives might provide another simpler option, if they could be networked together in a p2p fashion. The cops might be able to confiscate my webcam and pocketpc, but my recordings (and proof) would be elsewhere in the aether.

    --
    geeks are cats who dig a certain kind of cool
  16. Re:Not The First w/ The Idea by Fnkmaster · · Score: 2, Interesting

    And oddly, Simpson Garfinkel, another well-known technopundit, submitted a very similar idea (P2P backup service) as a business plan to the MIT 50k competition back in 2002. See here for the entry summary (search in the page for Garfinkel). Anyway, I somehow dredged that up from the back of my brain when I saw this Cringely piece because I recalled that Garfinkel was interested in actually doing something like this several years back.

  17. Re:Interesting idea by bloo9298 · · Score: 2, Interesting

    On the contrary, I'd say Auntie has a really strong case that she never had the key to someone else's encrypted data stored on her drive, so the RIP act would not apply to her.

  18. Re:p2p backup by aqua · · Score: 3, Interesting

    I made a related waggish proposal a couple of years ago:

    1. Make tarball of backup
    2. Encrypt if desired
    3. Encode tarball, 4-8 bytes at a time, in email addresses
    4. Put email addresses on web
    5. Wait for spam

    Presto -- spammers now pay for your backup; anytime you have a disk failure, just wait a while and watch your spamcan or smtp log, and reconstruct your backup at will.

    (Some assembly required, offer void where prohibited)

  19. Prior art by isomeme · · Score: 2, Interesting

    An equivalent idea was proposed in about 1982, at the dawn of the internet. Simply tar your filesystem, then email the tar to yourself along a lengthy old-style routing chain. If you need your data back, just wait for the email to arrive and untar it. You could tune the recovery latency by adjusting the routing chain. Of course, over dialup uucp, even one-node-out-and-back path could result in a two day latency.

    Man, those were the days.

    --
    When all you have is a hammer, everything looks like a skull.
  20. Re:No thanks by pHDNgell · · Score: 2, Interesting

    Maybe this would be good for some data, but I would never backup sensitive data on something like this. Nor would a lot of businesses.

    I've been backing up sensitive data almost exactly like this for quite a while now. I've got an application that breaks a stream of data into chunks and encrypts them. It compares the md5 of the source block against the md5 of the same block from the previous backup. If they match, it hard links the block into the backup directory, if they don't match, it encrypts the block and stores a new copy.

    The net effect is that I have about 2.5GB of full database dumps nightly that take up about 3MB of storage space (thus takes about 3MB to transfer it with rsync -H). I do this without having to store the unencrypted stream.

    So yeah, it may be better to not send my GPG encrypted blocks to these other machines, but I trust GPG about as much as I trust my own network and applications keeping my data safe, so it's good for me.

    I do the same thing with my DB dumps and my mail. I only have a 144kbps internet connection, so being able send my full dumps out nightly across it is a real win.

    --
    -- The world is watching America, and America is watching TV.