I uploaded a legit Office ISO I downloaded from MSDN to SkyDrive (as it was called back then) to test throughput up and down.
After 3-4 hours it was gone.
My data isn't going near it.
Edit: just to clarify... I think they hash incoming files and delete known ones as part of a takedown system. This was an ISO that had been shared on TPB as well as MSDN. Of course their policy allows these measures but I'm not happy putting something up there on the basis that they can arbitrarily delete it.
For copyrighted content, we have to respond to DMCA notices like other services. Sharing content to the public and getting reported by a third party is the only path for that. And in those cases, you definitely get a specific notice about the takedown. The web UI would also show you exactly which file was affected, and prevent you from sharing it again. It doesn't just delete files. (That would be unacceptable.)
Note that there's currently a 2 GB file size limit. It seems like the most likely explanation is that you put a large ISO in your SkyDrive folder, and it never succeeded in uploading because it exceeded the limit.
The file was 600Mbish. It successfully uploaded via the desktop client then was later downloaded on another machine via the desktop sync thing. Later that evening it was gone. I didn't delete it (it wasn't in the recycle bin) and I confirmed it was the correct live account I was signed in as. To be clear it was still on the source machine but not the destination. No antivirus had quarantined at either end.
That by elimination would suggest that either:
1. There is a reliability limit somewhere which is unknown and unpublished or a synchronisation bug.
2. You're unaware of a process or a false positive.
I should have opened another account to test this against with the same file but to be honest if I found a bug, windows live support has been abysmal. Hell they couldn't even work out how to close my account when the close account page refused to work...
FWIW, I can say with confidence that your issue had nothing to do with the fact that the file was an Office ISO.
How do you handle child porn filtering?
I have three children and take photos of them. I am worried about false positivies.
1. PhotoDNA runs on every upload. It's only used to identify known child pornography that has already been reported, to make sure it can't be re-uploaded.
2. Automated flesh tone detection only runs when a photo is shared. (This is a change in policy; it used to run on upload.) There are heuristics that try to measure whether it's personal sharing or broad sharing, and we're continually improving those. The goal is to make flesh tone detection only run during broad sharing.
3. If the broad sharing criteria is met and automated flesh tone detection triggers a positive result, that is the only case in which an item is anonymously sent to manual review. It's some highly controlled clean-room environment where a dedicated team tries to determine whether the content is a legal risk or not. Clear cases of shared child exploitation porn are reported. (A parent's "baby in bathtub" type of photos are not the target here.) In most cases, it's adult pornography or family photos. In those cases, the folder is marked as porn and simply can't be shared again. (There's a user-visible message on the web UI.) It's not deleted, and it continues to be fully accessible to the owner across all machines.
The scanning policy used to be more aggressive and didn't exclude content that was unshared or only shared to a small set of people. None of us liked that policy to begin with, and then some high-profile false positives helped force the policy to be revised.
However, the terms that are linked to me at the bottom of OneDrive.com specifically tell me that uploading porn is not allowed and presumably (haven;t double checked) tell me that if I do my MSA will be deactivated.
It's nice to have you and co. tell me that you allow porn, but the fact that the terms I legally agree to contradict what you say sort of puts me in an uncomfortable position.
I, personally, never use these file storage services to hold anything but encrypted data assuming that such data can disappear overnight without any warning.
If you roll your own solution, do you limit to desktop usage, or how do you get mobile support as well?
I have several ISOs from TechNet and they are all still there.
There are very few digital things that are illegal to possess (and no, a digital copy of The Matrix is not illegal to possess), and I believe Microsoft has both a legal and moral obligation to ensure the files on its servers are not child porn.
I don't think this measure is effective (it's trivial even for non-techies to circumvent hash-based detection) nor proportional to the likely incidence/impact of this crime. While I cannot speak to what their legal obligations are, I don't think we should encourage invasion of privacy for no particularly good reason. As such, I strongly object to the notion they have a moral obligation to commit this particular harm. Facetiously claiming that doing so somehow protects children is just a joke in particularly bad taste.
I was under the impression that PhotoDNA was more than simple hashing - that it used an algorithm like Google's "similar images"
But it is my belief that file hosters have a legal and moral obligation not to know what they are hosting for their customers. It is not the obligation of a host to look at a customer's files and decide what is ok and what is not. It is the same thing with Tor nodes or if you decide to encrypt files. I actually think that any online sync host like this or like Dropbox should not have any chance of knowing what I store with them. That's my private business and there is zero reason for them to know, is there?
I actually think that any online sync host like this or like Dropbox should not have any chance of knowing what I store with them
Cryptography is readily available to anyone who cares to use it.
2. You ought to provide some kind of legal citation for your claim that a judge can order a service like Onedrive to shut down after just two allegations of illegal content.
End-to-end cryptography should be the standard.
Really? It was my understanding that True crypt was done for and I don't know of any other similar software that is open source (ie trust worthy).
Although we all have reason to believe every single encryption tool out there is compromised, where does that leave us? Paralyzed. IMHO, we won't have 100% trustworthy software ever and we're better off using what we have then not using anything at all. Remember, in this thread we're talking about our privacy, not some top-secret project for which developing a encryption tool from scratch would be affordable (that'd be cool though).
Imagine having 10,000 files up there and they delete one...that would keep me up at night.
This OneDrive does not change metadata, it's a 'simple' cloud storage.
BTW: interesting, I've submitted this link an hour or so ago, and it was immediately flagged as dead. :)
Right now the best Cloud-storage desktop clients: Dropbox/Google Drive/Box/Copy/OneDrive, all do sync of the entire repository (unless you're willing to do the work of selective sync). Apart from that you're not left with many options, other than the web, for moving data to and from your giant 1TB account. ExpanDrive bridges that gap by providing access as a network drive. You connect and interact with the data without needing to bring it all in. We keep a big local cache, and do writes in the background so your saves feel instant, like Dropbox. If you've not checked out the software recently, take another look.
I know AWS is fine with this sort of usage (they charge for storage + data access) - any sense as to whether Dropbox/OneDrive/Google Drive are okay with applications being written on top of their cloud storage?
I.E. is this a hack, or something I should feel comfortable using as a storage option. And, other than AWS, are there any vendors you feel are most likely to be okay with using their service in this way?
What is the difference between ExpandDrive and Strongsync?
Strongsync is a hot folder, like Dropbox. Works directly atop S3 with no server needed. Full conflict resolution, etc.
ExpanDrive is a network drive.
I just got a download link and further down the page the price of a upgrade from a older version.
Before I download I am more interested in what I am downloading (is it a time limited demo?) and what the price is if I want a license. Otherwise its not worth my time to try the demo download.
Curious--any chance you can put together support for OneDrive for Business? Microsoft currently lacks a proper client for that service on the Mac, and it's a minor pain.
I also managed to bring down Explorer when opening Google Drive but I'll get a better bug report to you soon.
The only thing that makes me sad is that CBFS (what ExpanDrive uses as its backend) is really the only solution for doing filesystem stuff on Windows. Dokan just needs some love and a code signing certificate or two.
CBFS is good. We wrote our own for v1&2. Better to let somebody else focus on that and we can focus on the rest.
It sounds like ExpanDrive would probably solve this issue for me. (At least until Microsoft hopefully adds back support for using OneDrive with local accounts at some point.)
ExpanDrive's WebDAV performance has been kind of rocky, though. From what I can tell, it doesn't appear to support byte range seeking. To get to the middle of a 2 GB file, it tries to download the first 1 GB. Is that something addressed in the v4 update?
Coming in the fall in v4.5
Edit: to answer my own question, the web page says "Linux version coming mid-2014" :)
Photos are only scanned for nudity if they're broadly shared. The definition of "broadly" is changing. It's designed so that content isn't scanned if it's unshared or if it's only shared with a small number of people. The goal is just to make sure that people aren't using the service to host massive public porn collections.
Would you trust a sync service called "File Destroyer for Consumers"?
I'd like to exclude things like node_modules, build folders etc. It's gotten to a point where I have to disable Dropbox during development and let it catch up later otherwise it chokes up CPU at 100% doing it's hashing.
Any advice would be greatly appreciated!
Use git with one of the free providers like github or bitbucket. You will find it so much easier.
This complicates the state of a development folder. If Dropbox went down mid upload or you hibernated your computer before it had completely uploaded (it happens), you then work on another computer and that uploads some files, suddenly you are going to get a merge conflict which Dropbox cannot easily help you with.
Software development is not one file being edited, its a collection of files in a particular state that give it meaning. Dropbox is good, its gotten better but its still not the right tool for the job.
Git (or any DVCS) is good enough that you can commit and push every 5mins if you wanted. The point is that the state a a whole makes sense.
Plus what about when you start working with other developers, you cannot all work in Dropbox, you'll get merge conflicts every day. What about bug finding features like bisect? Or just having a log of when thing happened?
Sounds like the only way to do it right now is via the Seadrive client proposed below - will definitely move to the first major provider that offers this service.
You've got to love competition, but I wonder at what point Onedrive's race to the bottom in terms of pricing starts to impact Dropbox's valuation?
Other interesting pricing tidbits:
OneDrive will come with 15 GB for free (up from 7 GB). Office 365 Personal ($7/month) will come with 1 TB of OneDrive storage.
For the first time I'm rethinking my $8.25/month subscription to Dropbox for 100 GB, particularly as I only have about 10 GB of data...
Dropbox is in a really tough spot, because if they even cut their prices in half (which would still be significantly more expensive than OneDrive), their revenue would drop tremendously right before an IPO. My guess is that the primary options are to stay the course, go public ASAP, hope you don't get bled too much from Google/Microsoft and try to find more reasons for people to pay a premium for Dropbox, or to sell to Apple, which is the only big player without a meaningful competitor.
iCloud Drive?. The Apple buying Dropbox ship has sailed. Steve Jobs tried to buy them and Dropbox said no. Turns out it was a good choice on their part, but there is pretty much no chance of it happening now.
eh, I've looked into getting into this market, and the price I would feel comfortable with (at my scale... which is not large) would be around a penny per gigabyte per replication per month. 2 replications is about the minimum I'm comfortable with, so the $0.02 per gigabyte/month seems pretty reasonable to me, especially because the big players have access to some pretty dramatic economies of scale that I do not.
Hard drives are what, $150 per 4tb drive or so, to purchase? Figure that it's another $2000 for every 36 drives for a low-end motherboard and chassis; figure a 36 month life, to be conservative, and that's $5.50/month in capital. figure $200/month per kw usable. a hard drive is going to use 5-10 watts That's $1-$2/month per 4tb. so total cost is going to be around $1.88 per tb per month. Now, multiply by 2.2, as I'm going to have 2x replications and make each replication raid6, (10 disk stripes) so my cost is $4.14/month/tb If I'm charging $0.02 per month per gigabyte, that's $20/month per tb revenue. (Of course, this is all seagate gigabytes. It's more complex if you use GiB.)
That's plenty of margin. Now, this doesn't count over-subscription, and if you sell in 100gb blocks, not all of it is gonna get used.
Now, especially if you are letting users use this for more than just backups, you have per-account overheads like abuse handling.
But yeah, overall? from where I stand? the "race to the bottom" isn't even keeping up with hard drive prices. In a real "race to the bottom" someone at my scale wouldn't be able to make reasonable margins.
The biggest problem I see with the market is that most consumers don't need that much storage. I'm going to need a lot of customers to just fill my first 74 disk cluster. Then, if I let customers share the files, I'm going to have to deal with dmca bullshit, and I ain't doin' that for free.
Unless you're a Dropbox equity holder why would you care? As a consumer all I care about is cost and reliability, neither of which are impacted by their valuation.
I'm moving at the end of the year when my subscription period ends.
I was a bit late to the online storage game and at $2 per month for 100GB, signing for Google Drive was a no brainer over DropBox.
This is relevant to me for http://emailitin.com/ because with email attachments there's no content-length. Since I don't want to store attachments on my end (even temporarily) I'd rather just pipe them directly to OneDrive (as I do with GDrive and DropBox). With the current API limitation I have to pipe to a temporary file, stat the file, and then upload. Really annoying.
Either way, paying for the 100GB if you need the storage is a no-brainer. And OneDrive is still one of the few truly cross-platform storage services that's also ships natively with an OS. Now that Apple's offering additional cloud options on iPad, I could see myself picking one cloud provider and sticking with it ....
Can anyone with OneDrive exp chime in?