Dropbox rival Space Monkey puts 'cloud' in your house | Hacker News Reader

Dropbox rival Space Monkey puts 'cloud' in your house

news.cnet.com

52 pointsivoflipse59 comments

They give me half of a 2TB drive? Doesn't that mean that I'm getting only one copy of my data stored outside my house? So if a power surge fries my equipment, there's one copy of my data spread across a couple dozen other machines. I guess that's fine as long as no one but me experiences any failures...

It's possible that they're just overselling, and they expect most people to use only a couple hundred megs, which would allow lots of redundancy. If so, that seems like a really bad idea, since heavy usage of an on-site disk is likely (let's back up all these pirated movies!), and their buildout requires either signing up new customers or swapping devices, neither of which is quick or cheap.

I wish they had gone with At least 2 copies offsite - that was one of my first thoughts too.

Then I remember that I used to do backups at most once a year until I moved to the Mac and Time Machine.

(edit: clarify sentence)

How do they guarantee at least two offsite copies? That would require that they give everyone 1TB out of 3TB. Unless they're also storing on their own servers, in which case the consumer-housed device seems pointless.

And two copies is still pretty low for flaky infrastructure. They're basically setting up your data as a torrent cloud, except there are no seeds if you lose your equipment, so you're trying to download fragments from a bunch of other machines, and each fragment is only on a couple of machines.

What I meant was: I wish they had gone with at least 2 copies offsite. Sorry for the ambiguity.

And you're right - 2 copies would mean 1TB out of 3TB.

Ah, thanks for clarifying.

In my 25 years using computers, I have never once had any piece of equipment fried by a power surge. I have had things fail, sure - but never by a power surge.

You've never had lightning take out equipment? I envy you. Few things are worse than hearing a crack, having the lights go out, and then smelling burning circuit boards.

I am sure it's due to geography. Looks like ill be in Dallas for the next 2 years on a project - so my experience may change :)

Happened at work last month. Everything from the mini-fridge to the monitors.

I would suggest they count on most people not using the full 1TB. Not really unusual.

Maybe they're hoping compression helps a bit? Having said that, I reckon most people won't fill up their drives...

So, the company gives you a 2TB external hard drive for free, on the condition that you use 1TB to store your own data and 1TB to store copies of other people's encrypted data? Sounds interesting, until you get a huge bill for bandwidth overage. I have barely 100GB of bandwidth to spare every month (out of a total of 300GB/mo), so if that hard drive gets filled in less than 10 months, we've got a problem. They're effectively pushing their costs onto ISPs, who in turn will push their costs onto customers. Restoring backups is also going to be pretty slow if your data is stored on somebody's 10M/1M DSL line. They'd better market it in countries where everyone has 100M and no bandwidth caps.

It seems to be working ok for Spotify

Not to mention the obvious legal ramifications, even of encrypted data - in the UK, it's a crime not to give up the key to encrypted data, even if you claim to not have it (though I would hope in this case it's provable you really never did, and that would somehow help.)

It's not dissimilar to the legal ramifications of browsing HN, then. Look, encrypted data!

xQq84/+wpysSnUveVUX6592sHzGBptxysvtginJr2OFkDifXYtNCFg==

Yes, bandwidth is a concern, but it should be easy to let users choose bandwidth limits. The question then is whether the product is still useful if it takes 10/20/30 months to fill the hard drive. I suspect that there's a segment of the population for which it will.

ISPs pay for bandwidth leaving and entering their network. inter-network bandwidth is actually free for them, so it's very likely spacemonkey could prefer peering with other users on the same ISP.

That would certainly help, but there's a limit to same-ISP peering. For example, I use a relatively small ISP, so not many people are going to be using the same ISP that I do. Even if you use a larger ISP, you're screwed if even a single person dumps 100GB/mo of data on your SpaceMonkey drive.

Upstream bandwidth on DOCSIS causes congestion that requires capex (node splits) to fix, so it's definitely not free. See http://www.ietf.org/proceedings/74/slides/alto-1.pdf

It's funny, it's just as the world started to become more sane again by storing data centrally now we have a company that will bring us back to storing things locally and share it out on links that at the last mile are not implemented to be full duplex. The economics of this is just wrong, and hopefully it won't take off.

This has been shown not to work: http://people.cs.umass.edu/~emery/pubs/TFS.pdf (section 5.1) I wonder how they think they're getting around the upstream replication bottleneck.

Upvoted, but I skimmed 5.1 and missed the "shown not to work" part. Can you be more specific?

Whenever a device fails or goes offline, the data that was stored there needs to be re-replicated from the source. But figure 6 shows that even with a 2 Mbps "turbo" uplink you can only replicate ~25 GB; if you have more data than that, replicas will fail faster than you can recreate them.

Sounds like that's addressed by storing pieces of data from many sources in the non-local part of space monkey storage: even if each source only has a 256K uplink, multiply that by 100 or 1000 and you're in pretty good shape.

Probably by preferring storage nodes on faster connections.

Interesting idea, but this worries me:

When you sign up for Space Monkey, you get a physical networked hard disk with 1TB of storage free for you to use. It's actually a 2TB drive; the rest gets used by other people on the system.

So 1TB is filled with other people's data? What about security? Dropbox had a simple security issue that potentially leaked all users' data. But now, the data is on your drive. You could mirror the drive ('dd') and try to break into other user's information without anyone suspecting anything.

Or what happens if someone stores child pr0n on SpaceMonkey, and it gets mirrored to your drive? Legally, are you now on the hook for possessing the information since it's on your device?

Good question. The Space Monkey software mitigates these issues in two ways: First, all data is carefully encrypted before it ever leaves the user's Space Monkey device for replication across the network. This is obviously a key concern for all of our potential customers, so a lot of time has been spent getting this part right.

Second, the system breaks each file into many small blocks with built-in parity and other such redundancy measures. These blocks are then spread across many devices to both increase redundancy of the network and distribute the bandwidth usage fairly across nodes. As a result, you wouldn't really ever store any identifiable data from any one other customer.

Therefore, if the government wanted to seize evidence, they would gain nothing useful by seizing any Space Monkey device other than the suspect's, which contains the local cache of the user's own data.

This is one of the most straightforward possible applications for cryptography in data storage- no worries about key negotiation or cipher negotiation or PKI infrastructures. And yes, they're encrypting your data before sending it anywhere- that's pretty explicitly mentioned everywhere the service is explained.

So, if you're asking whether you're legally liable for the content of data on your hard drive which you have no possibility of decrypting, um, the answer is "no".

Depending on the jurisdiction, you might still be liable for other people's encrypted data stored in your property. Whether anyone can prove that you have any particular piece of incriminating data is a different story of course, but law enforcement agencies in certain countries have a nasty tendency to skip proofs.

I submit that you're probably screwed anyway in those countries and/or jurisdictions, whether or not you have someone else's unreadable contraband. :)

In fact, here's an exercise. The following data represents a base64-encoded encrypted image:

qipN5pKHq+CCBm+yfHz0Y/NmkLd+1GjuowopHYWs3qjo6wuPsrulOBZJUXGt tsc465cM9p2qUZvKYxt/5M+2+zkjdbNMi447cLY4KUw3Xb934KDkyrviZYkh gryO3C5cPw0ZVitr/6e3pyQiEGga4SwbskCQhMw7XlqDYJowpCs/Qsh8ZQiR 2/+zQ/CGhYfU37dIQYVzqPCM7y5DHcDePGvzx2BCzu8OQThLnMWGiz5kODzY 49zrfoTQg8W0NKIC7rD529CnPVyfQok+u8L8LGqZSbOh2CUygWIF3mjmTGtd EnUBrkI49aVjeEa9/anCUw2nBRbmWWZZ5p1bvDYHGkEtlaFTJrSfVnuejBa1 NqNdMFIU6qAlpJ75Js8O2+jNi+qbVE/JUjYFtB45mf0B/E4GXbBcjg/gWhSi LvTOX7P8+HDMhg8jB2nnJ2Yf3l2CPopjKczWGZMUu3P8jamdp2kv67q0kIrE xGc3sr8r8R78StuQUVpp+5SexykhDOMQGowdBQ2yLYtRtoNDlWm47x7agM8i YP70n4wLhwZ03JH+2WmmVLVx2Yk4AZIU4t7XfdaoYaC33U58QeSaBJ4fdsJO PBp1L559jJTRrBk996gLN7V0oQSRMCWF67KezK3Bk2a07vzxjiXfwJOB5BDl ooSvoTdu5Slv9C+aIoa4ade7r7umC8Ac84XCfBY/UlmWIq0VogXuzOzZ2WkX 0uUpCnz2WX3rShKFFd8w9HhoxsOaR2S3sFbL7foqpx9SXYYjpM2vJHpEjMFQ W7Xmqlvb/2m1aUB2f6qFmqKPjo5/zIAOl90YDoDaQSin9C9/X6WRydHHcu1/ mrjo5LSdfouqOwVXMDnh95/YPvSGdO8GP35pRKUceidM4oYyQCRYGb3xKn0O TP/jKy3MXz/CeZsbxQtobq1DlC6NVVp5UifQM72ufdI5xcqObQSECFXKTTtt FuGdg+ZQ/0yLQ6WQde1B44LJML0MWIYPjrO74tOiCn4VZEsC9lakMMJZ2Qyp M1BRsh7xeQ0eVMOCEe5+yDV45uvchoGyAZsr2qTcgqTCKm4U9r1PUqBjWUjS AkVE3uIHjwL7D+gDvRuvNk4PdHpUX0p+oLp2ATmcR+TmhSNL8UEuAAw1ajgH SlhtS/BdXWkC5SFbfp6Dvt+WD+/2CwAZGgImAWXPv0YM+za80Q99au4NPR+L +KQW2wjChNkEKau9Jq5uGf36Vnt08dxVRoWzsWcPHmlFquXQrsBVsbe2Yu6v EfTAwaxFUB1kgLdTWJlxObfiBtzwf855yIggNcJZ9b/e9k6NpdJbCBYQJElf RN5qYGh4A6ASpZgZa86e2nPHvIJprm7IN9uqiI2EP9fPQp+zLorNKwxnZkv7 W//D28p0Wy5tS5F5p2ee7QI=

Is it pr0n? Is it an image of US gov't classified data? Is it hate speech? Who knows, but now it's on your computer. There are a lot of other ways someone could get data onto your computer without you knowing if it has a particular meaning.

This is a great idea for the long term. There will always be a need for something like this, as an alternative to the cloud. Dropbox should have come up with this, to differentiate themselves from their competitors.

Sounds like a good acquisition candidate.

Actually, you know who could probably sell this? Comcast. Make a single device that serves as a cable modem and also has this storage angle built in as either a "free" service with the bundle or as a small upsell.

I think Comcast wants to reduce upstream congestion, not increase it.

Edit: I'm talking about upstream on the last mile.

Why? If anything, I'd expect them to have plenty of upstream bandwidth, given that typical traffic patterns are primarily downstream.

They sell the upstream to hosting providers. http://en.wikipedia.org/wiki/Peering

I don't see how this is relevant. If Comcast were to sell this (or an equivalent device), they'd keep all the traffic within their own network.

When you have a peering contract, you offer your own network. Every bit not used is so valuable for cable providers, we consumers forget that because we all have flatrates.

I don't see how this is valuable. By definition a peering contract is not staying inside Comcast's network. The bandwidth for this (if Comcast were selling it) would stay entirely within the network, where it's not significantly competing with peered traffic.

Maybe I'm misunderstanding what you're suggesting.

I would love to see a Time machine equivalent for Windows. I already have a 2TB storage device at home but pretty much every local backup/sync service sucks cojones. I'd gladly pay a flat fee for a product that does this

I think this is a great idea but I think Dropbox has a pretty well established brand name. My guess is that most people aren't going to find the need to switch. The ones who might make the switch are the paranoid folks who would think this is some sort of more secure way to sync.

The idea is nice but the challenge, IMO, will be in marketing this.

Dropbox is also kinda expensive. $125/year per user starts to add up if you've got a lot of users.

How did you arrive at $125/user? A link for the unaware?

$125/user is the amount it costs to add another user to a Dropbox for Teams account after your first five users.

https://www2.dropbox.com/teams/pricing

It depends on your financial situation in life. Instead of thinking "expensive" versus "cheap" -- because they're just words with no clear definition -- you need to think about the cost for what you are getting.

$125/year is just over $10/month. Nowadays in the US/UK/Canada, that's about 2 cups of coffee at Starbucks.

For that price, you are receiving 1TB of backup space for yourself.

Each person is $10/month, which is useful if you need to collaborate and not worry about setup.

Insurance on your home always varies, but usually runs about $50-100 per month. And that's just the price you pay for when things are going well.

For me, there's a cost to peace of mind and the ability for me to save time in my life to do things other than maintenance.

FWIW, I look at Dropbox as a monthly insurance premium to ensure that my files will always be around and I don't need to worry that my in-home RAID server must blow up at any time.

I was actually thinking of B2B. $125/user isn't a lot for one person, but it is for 50. And I would imagine most businesses have some (worse) system for sharing files already...

Didn't Wuala use to do something like this? Except more for P2P file backup rather than sync?

P2P backup / sync has been done before (allmydata is another example from longer ago). It's that plus the appliance that makes Space Monkey unique, IMO.

IIRC what wuala did was give people more storage if they give up a part of their hard drive for other to sync with, so it's exactly the same model except without external hardware and more flexibility in choosing how much storage you want to trade off. Wuala also keep a centralize server in addition to this P2P cloud so I think the purpose of the cloud was more for bandwidth.

it's one thing to pay money and gain space, but it's another thing to pay money and lose disk space you have. i feel like people are more likely to pick the former.

This sounds awesome! I've been waiting to see a step this direction for a long time. For a lot of people dropbox is the best tool for files, but its ridiculous for the trip between two computers in the same room to have a layover in A3.

That's why dropbox has lan sync: https://www.dropbox.com/help/137

That has a few big catches, including maintaining an internet connection and being in the same broadcast domain. And looks like it doesn't have lan-only drives, I maintain a box serving up samba shares specifically because the work files in a specific project work files would blow my bandwidth cap for the month in 2 round trips.

> That has a few big catches, including maintaining an internet connection and being in the same broadcast domain.

Are those actual big catches? Those seem like pretty reasonable constraints. What machine do you have attached to your LAN that can't access the internet? And the network restriction is probably just fine for >99% of their customers.

I know its not going to impact the vast majority of their users, so they don't have to care. For my uses, they're all pretty big. That's why I'm so happy to see a new product that may fit into those situations better. Not every computer is on a SOHO network, or a network with many properties of the same.

The network I'm on most of the time is huge which means its heavily subnetted and no one user gets great throughput. My apartment has the aforementioned cap. My old man's place has no internet connection at all, just wifi for getting files to the media computer connected to the TV. By building a product that can serve these users, they'll have a product that can better serve users on ideal connections too!

I don't really see how this product makes any sense for the scenarios you're describing. This system still needs to be able to hit the Internet at reasonable speed, or else you get no offsite backup. It just becomes a network-attached hard drive, and if that's what you want, you can buy one at a better price with more storage (and it won't waste your bandwidth trying to be a "cloud" for your data or anyone else's data).

My original post was that this is a step in the right direction. From one thing I happened to mention, this thread took a tangent about dropbox's lan sync, but thats not specifically or only the area I care about. There is a place I'd like to see file storage tech to go, and we are inching towards it, but its not delivered with an off the shelf NAS box, dropbox or simple combination of the two.

I guess I'm not clear what your ideal product would be. All of the scenarios you describe have very constrained bandwidth, which means that sync outside the LAN will always be crappy. It sounds like you either need NAS or peer-to-peer sync (something like Mesh or AeroFS).

I'm glad to see Dropbox as defining this space.

And here I thought it was an actually solid rival to dropbox you could install in your own servers and enjoy the same level of functionaly.

Yes I know there are some hacks around using FTP apps to get a similar experience, but is not the same, is still clunky, sometimes too buggy, and let alone the average dropbox user wouldn't be able to even read the instructions, let alone put it together.

Reply on news.ycombinator.com