
A fool and his data
are easily parted on a remote cloud.
Amazon.com has become the latest outfit to decide copying Dropbox is a fine idea, further cloudifying its Cloud Drive with a desktop client that does the usual store 'n' synch between multiple devices while also offering some interesting insights into cloud economics. The nominally-profitable online retailer has operated Cloud …
Having used dropbox, sugarsync and skydrive - I noticed that they will upload multi-gigabyte ISO's in a single second over my shitty ADSL connection - clearly they are hashing files for reference and storing common files once. When you consider how many people are storing mp3's, iso's and avi's that have dupes you can deduce that their total required capacity on the wholesale side of the equation is substantially less than the cumulative amount of customer data they take in.
Maybe... or maybe it just uploads them in the background.
It will take a lot more than a second even to calculate a local hash of your "multi-gigabyte" file. Reading the file will probably run at 50-300MB/sec (HD-SSD), but even if the whole file were in RAM cache the CPU time is a limiting factor.
For example, to hash (SHA1) a 2GB stream of zeros on a Macbook Air core i7 1.8GHz takes 10 seconds of CPU time:
$ dd if=/dev/zero bs=1024k count=2048 | time shasum
md5 is about 7 seconds, but that would be a risky choice given the ease of generating files with the same md5.
Anyway, you can prove it easily one way or the other: encrypt your ISO using something like gpg and a random symmetric passphrase, and see how long it takes to upload.
"Or he could just look at his bandwidth meter."
Every time I've tried these cloudy backup data services the bandwidth graph pegs itself - pun intended. Stuffing data up the pipe just results in crap performance at some level no matter how you queue the data - web pages that load as fast as a 5.25 SSSD floppy.
DropBox used to do that (hash the local content, send the hash, if it matches the file is assumed to be 'yours' and copied from someone else.
They had to stop once people realised how to fake the API calls, b/c it allowed unlimited copying of, well, anything...
This post has been deleted by its author
I discovered OwnCloud a few months ago and set it up as a front end to my S3 storage, having similar visions of owning my own 'cheap-as-chips' cloud. I dunno if it's improved any since then, but I found it less than beta quality and not fit for purpose. I had a similar experience with my 'cunning plan' to replace my Flickr account with OpenPhoto.
The unfortunate thing is that a lot of OSS projects of this kind are borne aloft on a wave of enthusiasm from folks like us who want them to be great, so we can 'sock it to the corporate man'. Sadly, it often turns out that 'the man' got to be so 'corporate' by doing it right in the first place and the roll-your-own alternatives just aren't up to scratch.
It's a good point but then you have to take into account the additional costs. Ok your time might not be worth anything, but maintenance should be taken into account. Electricity on your server + router is also going to cost. Say 24/7 = 730hours a month running at 50watts(*) at a UK average of 14p per unit is £5.11 per month. Which is a similar price to the article, (but you get to have more space if required - still effects running cost). Security is down to you. Obviously with the off the shelf option you assume they know what they are doing, but if paranoid you just true crypt an image. However security also relies on the reliability of your access as well. No point in having your own personal server if BT Openworld are doing their random ''lets cut everyone off at the exchange thing. - (mind the same argument applies for the external server provider too). Just a thought.
(*) In my experience, people(**) who do this setup actually use some old crappy computer, (which is suitable to serve), but ignore the fact that that old crappy computer is actually sucking more.
(**) Me as well, but it was fun/learning experience. Just don't think it's really any better than having an external solution and your own local backups.
I always wonder why 'Da Reg', doesn't have a facility to list up and down voters by 'name', (obviously Anon's can be disregarded because their views are rubbish). But if someone goes to the trouble to neg something, it might be nice knowing what the counter case is, or who actually might of made it.
I can think of several counter arguments in my less the eloquent post, but the down neg(s), are a really bad tool to show the errors of my ways. Throwing a brick is easy, going up to someone, explaining the function of the brick, and then shoving it their face, (while all they're peers are around), less so.
This post has been deleted by its author
Any data in it has to be retrieved or will be lost forever. Wonderful. Does anyone really think that cloud storage is even slightly dependable?
The 'replacement' for digital vault is only available if you have BT Broadband. And even then "This service will be stopped and your back-up files deleted if you don't use it for 90 days." . WHAT KIND OF CRETINS ARE BT LOOKING FOR?