6 ms·

The dropbox endgame

21 points by ZeroMinx 16y ago

edo 16y ago
Apple already made the first step by releasing an OS that is geared around the idea of cloud-based computing (Lion). It's just a matter of time before this happens, and when it does; Dropbox might have a hard(er) time competing. http://www.patentlyapple.com/patently-apple/2011/02/apple-in-the-sky-with-diamonds-a-cloud-based-safe-deposit-box.html#more http://www.patentlyapple.com/patently-apple/2011/02/apple-in...
- SkyMarshal 16y ago
  Apple's market share is still relatively low, a bigger risk is if MS tries build something like this into Windows too.
malux85 16y ago
I'm not trolling - but this is hardly a 'weird idea' as the article states ... I thought of this and did it last weekend :) My flatmates harddisk broke down the other day in his laptop, so I put Ubuntu live CD in, a USB stick and hooked up dropbox. A script later, and he's using dropbox for permanent storage online, and considering not buying another hard disk. Of course he will :P .. but my point is that this seems a pretty obvious progression :)
kirchhoff 16y ago
Doesn't seem wise to entirely trust your data to a single entity. Can we be sure dropbox will still exist in 5, 10 years time? What happens when your net connection fails? Local storage will forever be around, in some form.
- Swannie 16y ago
  Millions of people don't appear to share your concern when it comes to their personal email accounts, which are arguably more valuable than a few documents... Actually the majority of computer users have no form of backup, despite how easy it should be. Lose your laptop/iPhone/hdd crash, and many people find themselves lost with no backup. Of course local storage will be around, in some form. But my guess is that that form will just be a cache, and for some people, a mirror of everything on a small box (NAS/plugpc type device) at home, which they can take everywhere for convenience.
  - kirchhoff 16y ago
    Fair point, although with email we're talking (for the majority of users) about huge companies which will almost certainly linger.
    - tybris 16y ago
      It seems like Dropbox is going to be a pretty big company, and their infrastructure largely runs on Amazon.com, which is a company with more revenue than Google.
uberalex 16y ago
This article is very interesting. The iTunes example in particular shows that these measures have to be transparent or they won't work very well. I think this was very much Microsoft's vision for Livemesh; they were thinking about desktop apps running data across the cloud. There's a real potential for dropbox if they could somehow hook into a variety of web apps and provide a common back end. The problem at the moment is that if you assume much of the app work people do will move into the browser, you're losing the common file base. You have several app-specific silos across the web.
timinman 16y ago
This is how I use Google. My docs, calendar, and email are everywhere at once - no syncing necessary.
EwanToo 16y ago
I'm not sure I agree at all with Jacques maths, the cost of local storage is tiny compared to the cost of Dropbox itself, and that's unlikely to ever change A hard drive costs maybe $30 for 100GB, higher capacities have much better GB/$ ratios. Dropbox costs $240 per year for 100GB, which has to cover both the storage and bandwidth costs. At the moment, the growth in data stored is increasing at much higher rate than the rate at which the cost of storage and bandwidth is decreasing. Sure Dropbox is great and a local/remote sync process is likely to become normal over time, but it won't be for cost saving reasons.
- Swannie 16y ago
  But most of that data stored is things like illegal music and film downloads.[* ] With the rise in popularity of Netflix, Spotify, Pandora, etc. people will only use local storage for a cache. Why would you waste time downloading it, when it's on tap? * Media producers are a little different in this respect and are always going to want local storage. However Jacques is mostly talking about the 95% of users who just have few 100mb of docs.
  - EwanToo 16y ago
    There's a lot of people buying legitimate music and films now you know..? Plus photos and videos that people make themselves will always take up a significant amount of space. I don't dispute that remote storage and streaming will become normal - I just think anyone who looks at the costs and thinks it'll be cheaper than having it locally stored is kidding themselves. It'll definitely be better than having it locally stored, but it'll cost more too.
planb 16y ago
I believe this would mean giving up the main advantage DropBox has over all competitors: It's the first cloud storage provider that sucessfully mixes local storage and cloud storage by providing transparent synchonization. After all, the DropBox folder is just a plain folder, but it is sync'ed whenever something changes. Without the local storage part there would be no reason to use DropBox as one could simply use for example a WebDAV share or SFTP.
k7d 16y ago
I'm pretty sure this is also the current vision of the Dropbox team. They are making a lot of effort to become a platform not just a product.
orionlogic 16y ago
Dropbox sits in that transition period from local to cloud storage. I said earlier in my comments that cloud will be an OS feature and Apple will lead that. However it will not start with pure storage(their approach of experience rather than specs). It will start with your music,video, pictures. Than your ios+local apps (with preferences and save files). And then Computer setup (remember your OS setup). Thats why Apple is easy with offering SSD harddisks in their top tier products.
- tybris 16y ago
  There's nothing transitional about it. This is what cloud storage looks like at its finest. Any model where you don't keep your files in local storage and sync is going to be slow, error-prone and annoying, just like all the solutions we've had before. There's certainly a case for more OS integration, and more powerful syncing, but this is the right model. I don't expect Apple to make such a big dent. They'll be in their Apple-only tunnel vision and would not be able to compete with a solution that can sync across Windows, Linux, Android, etc.
  - orionlogic 16y ago
    By transition i mean being more than storage solutions. Its computational network power, location independance,security etc. all the things related to cloud. You remember SETI project? Wouldn't be great if something is already integrated within. I don't see any bad in company-only tunnel visions as long as it drives innovation.
mikecane 16y ago
The thing I don't like about cloud services is this: They can be served a search warrant to let government go through what belongs to me without my ever being notified.
- dan00 16y ago
  Well, put a magic folder into your magic folder: encfs. ;)
- EwanToo 16y ago
  I think Dropbox claim they can't see the files you stored unless you share it with someone else or copy it into the public folder. I think the rest is encrypted using a key generated against your password and they only store the hashed version of your password, which couldn't be used to generate the key.
  - drdaeman 16y ago
    I'd call shenanigans. Typical password gives entropy that is way too low for any reasonable security.
beza1e1 16y ago
Does the author mistake the past with the future? A cloud-based home dir is basically a network file system (samba,afs,nfs,...) We had them for decades and they suck. Dropbox is an improvement, because is not necessary anymore to be "always on".
- tybris 16y ago
  The difference is Dropbox has some 21st century common sense that actually make it a viable solution for the cloud: Eventual consistency is good enough but I agree, many of the points he's making are nothing specific to Dropbox.
Sapient 16y ago
How will this work out in a world where the telcos seem to be bent on increasing costs and introducing data caps? I live in South Africa, where the normal ADSL account gives you 3 Gb of traffic a month (10USD/gig for each gig after that) at a maximum of 4 Mbit/s (10 in a few areas). I know we are probably an outlier, but judging by what happened in Canada recently, and a few cases in the US, its seems possible a reverse case of Moore's law will come into play. Also, most people (speaking globally), wont have the network capacity to support this kind of system for many years. There would likely have to be a truly massive increase in network capacity before we ever see this really take off. For fun, try running your computer off an NFS - even with a 100Mbit connection, it gets annoying quickly.
DrJokepu 16y ago
As long as there's no 3G or other (affordable) wireless internet access on the London Underground or anywhere else where I would need my files, I'm simply not interested in purchasing a product that doesn't have local storage. I mean, imagine an iPhone that would stream everything, I couldn't even listen to music during my daily commute.
hemancuso 16y ago
Sounds like the Chrome OS vision to me. Dropbox is a great place to store media and created content but doesn't easily accommodate your registry hive or myriad support directories that you need to make programs actually run. If you re-architect to achieve all of those goals and still have sync, you get ChromeOS. Barring that you have a nice replicated USB key (which Drew describes as the original original vision)
srainier 16y ago
Dropbox's core competency isn't cloud storage, it's continuous client syncing. If anything, moving towards not using the local file system at all hurts Dropbox, as they lose the part of their service that makes them special - syncing all files locally. On top of that, I can't imagine removing a local file system would work as a general purpose solution. By the time network bandwidth catches up to today's bus speeds (probably years away), the new bus speeds will be that much faster, and that's what people will want. Managing files locally may be a pain, but that's a UI issue. Users want native performance. Cloud files are a leaky abstraction. Dropbox gives you local files and abstracts away cloud storage and syncing. That's the way to do it.
GrandMasterBirt 16y ago
I in a way agree with the OP however I think we are decades away from the proposed future. The problem is multi-tiered - If you are outside of a major city, chances are you don't have internet wireless coverage everywhere. In fact many places in the city face this issue. This is more of a physics issue than anything. Inside my office building I get fairly bad coverage. Don't forget tunnels and such. My apartment has ZERO coverage. Only my wifi. And not all apartments I visit have wifi or coverage. - Price and transfer rate. Sending my hard drive to dropbox is about $50 a month (or more) and not to mention taking about a month to actually upload. - I don't want to rely on a company like Dropbox to exist. If they fail (see delicious) I don't want to be stuck with nothing. In fact single points of failure are bad. And I don't mean Dropbox can have a hard drive failure, I mean what if DB decides to triple their price. - America's internet infrastructure kinda sucks with current insanity pushed by isps regarding data caps/etc. Just wait till people REALLY put our infrastructure to the test.
orijing 16y ago
Very interesting idea. However, it made me think that one major flaw of Dropbox will limit its potential as the "ultimate storage." Dropbox is very memory intensive, even for my 5 gigs and my 40,000 files in the Box. If you cover my entire hard drive, that's over 100 gigs and millions of files. Each file (and each sub part of a file, sort like like a Merkel Tree) must have its finger print stored in memory. Currently those 40,000 files use about 70 MB, and when syncing, consume an entire CPU. I can't imagine how resource-intensive it would become if I indexed my entire hard drive now. I don't think Dropbox can push order-of-magnitude improvements in resource usage, even if they move away from Python. I'm sure they're already doing very well. I have multiple friends working there, and I have talked to Drew and Arash. All of them are very brilliant people and I doubt that they haven't consumed all the low hanging fruit in performance. Sure, in the future using 20x the memory may be OK. That's just 4.3 doublings, so maybe memory increases may balance it out. But don't forget that the amount of data I have will also increase. So I don't think it's feasible (with the current trajectory of technology) to have Dropbox serve as the ultimate storage medium, unless we develop a different way of syncing files.

21 points by ZeroMinx 16y ago

edo 16y ago

Apple already made the first step by releasing an OS that is geared around the idea of cloud-based computing (Lion). It's just a matter of time before this happens, and when it does; Dropbox might have a hard(er) time competing. http://www.patentlyapple.com/patently-apple/2011/02/apple-in-the-sky-with-diamonds-a-cloud-based-safe-deposit-box.html#more http://www.patentlyapple.com/patently-apple/2011/02/apple-in...

SkyMarshal 16y ago
Apple's market share is still relatively low, a bigger risk is if MS tries build something like this into Windows too.

malux85 16y ago

I'm not trolling - but this is hardly a 'weird idea' as the article states ... I thought of this and did it last weekend :) My flatmates harddisk broke down the other day in his laptop, so I put Ubuntu live CD in, a USB stick and hooked up dropbox. A script later, and he's using dropbox for permanent storage online, and considering not buying another hard disk. Of course he will :P .. but my point is that this seems a pretty obvious progression :)

kirchhoff 16y ago

Doesn't seem wise to entirely trust your data to a single entity. Can we be sure dropbox will still exist in 5, 10 years time? What happens when your net connection fails? Local storage will forever be around, in some form.

Swannie 16y ago
Millions of people don't appear to share your concern when it comes to their personal email accounts, which are arguably more valuable than a few documents... Actually the majority of computer users have no form of backup, despite how easy it should be. Lose your laptop/iPhone/hdd crash, and many people find themselves lost with no backup. Of course local storage will be around, in some form. But my guess is that that form will just be a cache, and for some people, a mirror of everything on a small box (NAS/plugpc type device) at home, which they can take everywhere for convenience.
- kirchhoff 16y ago
  Fair point, although with email we're talking (for the majority of users) about huge companies which will almost certainly linger.
  - tybris 16y ago
    It seems like Dropbox is going to be a pretty big company, and their infrastructure largely runs on Amazon.com, which is a company with more revenue than Google.

uberalex 16y ago

This article is very interesting. The iTunes example in particular shows that these measures have to be transparent or they won't work very well. I think this was very much Microsoft's vision for Livemesh; they were thinking about desktop apps running data across the cloud. There's a real potential for dropbox if they could somehow hook into a variety of web apps and provide a common back end. The problem at the moment is that if you assume much of the app work people do will move into the browser, you're losing the common file base. You have several app-specific silos across the web.

timinman 16y ago

This is how I use Google. My docs, calendar, and email are everywhere at once - no syncing necessary.

EwanToo 16y ago

I'm not sure I agree at all with Jacques maths, the cost of local storage is tiny compared to the cost of Dropbox itself, and that's unlikely to ever change A hard drive costs maybe $30 for 100GB, higher capacities have much better GB/$ ratios. Dropbox costs $240 per year for 100GB, which has to cover both the storage and bandwidth costs. At the moment, the growth in data stored is increasing at much higher rate than the rate at which the cost of storage and bandwidth is decreasing. Sure Dropbox is great and a local/remote sync process is likely to become normal over time, but it won't be for cost saving reasons.

Swannie 16y ago
But most of that data stored is things like illegal music and film downloads.[* ] With the rise in popularity of Netflix, Spotify, Pandora, etc. people will only use local storage for a cache. Why would you waste time downloading it, when it's on tap? * Media producers are a little different in this respect and are always going to want local storage. However Jacques is mostly talking about the 95% of users who just have few 100mb of docs.
- EwanToo 16y ago
  There's a lot of people buying legitimate music and films now you know..? Plus photos and videos that people make themselves will always take up a significant amount of space. I don't dispute that remote storage and streaming will become normal - I just think anyone who looks at the costs and thinks it'll be cheaper than having it locally stored is kidding themselves. It'll definitely be better than having it locally stored, but it'll cost more too.

planb 16y ago

I believe this would mean giving up the main advantage DropBox has over all competitors: It's the first cloud storage provider that sucessfully mixes local storage and cloud storage by providing transparent synchonization. After all, the DropBox folder is just a plain folder, but it is sync'ed whenever something changes. Without the local storage part there would be no reason to use DropBox as one could simply use for example a WebDAV share or SFTP.

k7d 16y ago

I'm pretty sure this is also the current vision of the Dropbox team. They are making a lot of effort to become a platform not just a product.

orionlogic 16y ago

Dropbox sits in that transition period from local to cloud storage. I said earlier in my comments that cloud will be an OS feature and Apple will lead that. However it will not start with pure storage(their approach of experience rather than specs). It will start with your music,video, pictures. Than your ios+local apps (with preferences and save files). And then Computer setup (remember your OS setup). Thats why Apple is easy with offering SSD harddisks in their top tier products.

tybris 16y ago
There's nothing transitional about it. This is what cloud storage looks like at its finest. Any model where you don't keep your files in local storage and sync is going to be slow, error-prone and annoying, just like all the solutions we've had before. There's certainly a case for more OS integration, and more powerful syncing, but this is the right model. I don't expect Apple to make such a big dent. They'll be in their Apple-only tunnel vision and would not be able to compete with a solution that can sync across Windows, Linux, Android, etc.
- orionlogic 16y ago
  By transition i mean being more than storage solutions. Its computational network power, location independance,security etc. all the things related to cloud. You remember SETI project? Wouldn't be great if something is already integrated within. I don't see any bad in company-only tunnel visions as long as it drives innovation.

mikecane 16y ago

The thing I don't like about cloud services is this: They can be served a search warrant to let government go through what belongs to me without my ever being notified.

dan00 16y ago
Well, put a magic folder into your magic folder: encfs. ;)
EwanToo 16y ago
I think Dropbox claim they can't see the files you stored unless you share it with someone else or copy it into the public folder. I think the rest is encrypted using a key generated against your password and they only store the hashed version of your password, which couldn't be used to generate the key.
- drdaeman 16y ago
  I'd call shenanigans. Typical password gives entropy that is way too low for any reasonable security.

beza1e1 16y ago

Does the author mistake the past with the future? A cloud-based home dir is basically a network file system (samba,afs,nfs,...) We had them for decades and they suck. Dropbox is an improvement, because is not necessary anymore to be "always on".

tybris 16y ago
The difference is Dropbox has some 21st century common sense that actually make it a viable solution for the cloud: Eventual consistency is good enough but I agree, many of the points he's making are nothing specific to Dropbox.

Sapient 16y ago

How will this work out in a world where the telcos seem to be bent on increasing costs and introducing data caps? I live in South Africa, where the normal ADSL account gives you 3 Gb of traffic a month (10USD/gig for each gig after that) at a maximum of 4 Mbit/s (10 in a few areas). I know we are probably an outlier, but judging by what happened in Canada recently, and a few cases in the US, its seems possible a reverse case of Moore's law will come into play. Also, most people (speaking globally), wont have the network capacity to support this kind of system for many years. There would likely have to be a truly massive increase in network capacity before we ever see this really take off. For fun, try running your computer off an NFS - even with a 100Mbit connection, it gets annoying quickly.

DrJokepu 16y ago

As long as there's no 3G or other (affordable) wireless internet access on the London Underground or anywhere else where I would need my files, I'm simply not interested in purchasing a product that doesn't have local storage. I mean, imagine an iPhone that would stream everything, I couldn't even listen to music during my daily commute.

hemancuso 16y ago

Sounds like the Chrome OS vision to me. Dropbox is a great place to store media and created content but doesn't easily accommodate your registry hive or myriad support directories that you need to make programs actually run. If you re-architect to achieve all of those goals and still have sync, you get ChromeOS. Barring that you have a nice replicated USB key (which Drew describes as the original original vision)

srainier 16y ago

Dropbox's core competency isn't cloud storage, it's continuous client syncing. If anything, moving towards not using the local file system at all hurts Dropbox, as they lose the part of their service that makes them special - syncing all files locally. On top of that, I can't imagine removing a local file system would work as a general purpose solution. By the time network bandwidth catches up to today's bus speeds (probably years away), the new bus speeds will be that much faster, and that's what people will want. Managing files locally may be a pain, but that's a UI issue. Users want native performance. Cloud files are a leaky abstraction. Dropbox gives you local files and abstracts away cloud storage and syncing. That's the way to do it.

GrandMasterBirt 16y ago

I in a way agree with the OP however I think we are decades away from the proposed future. The problem is multi-tiered - If you are outside of a major city, chances are you don't have internet wireless coverage everywhere. In fact many places in the city face this issue. This is more of a physics issue than anything. Inside my office building I get fairly bad coverage. Don't forget tunnels and such. My apartment has ZERO coverage. Only my wifi. And not all apartments I visit have wifi or coverage. - Price and transfer rate. Sending my hard drive to dropbox is about $50 a month (or more) and not to mention taking about a month to actually upload. - I don't want to rely on a company like Dropbox to exist. If they fail (see delicious) I don't want to be stuck with nothing. In fact single points of failure are bad. And I don't mean Dropbox can have a hard drive failure, I mean what if DB decides to triple their price. - America's internet infrastructure kinda sucks with current insanity pushed by isps regarding data caps/etc. Just wait till people REALLY put our infrastructure to the test.

orijing 16y ago

Very interesting idea. However, it made me think that one major flaw of Dropbox will limit its potential as the "ultimate storage." Dropbox is very memory intensive, even for my 5 gigs and my 40,000 files in the Box. If you cover my entire hard drive, that's over 100 gigs and millions of files. Each file (and each sub part of a file, sort like like a Merkel Tree) must have its finger print stored in memory. Currently those 40,000 files use about 70 MB, and when syncing, consume an entire CPU. I can't imagine how resource-intensive it would become if I indexed my entire hard drive now. I don't think Dropbox can push order-of-magnitude improvements in resource usage, even if they move away from Python. I'm sure they're already doing very well. I have multiple friends working there, and I have talked to Drew and Arash. All of them are very brilliant people and I doubt that they haven't consumed all the low hanging fruit in performance. Sure, in the future using 20x the memory may be OK. That's just 4.3 doublings, so maybe memory increases may balance it out. But don't forget that the amount of data I have will also increase. So I don't think it's feasible (with the current trajectory of technology) to have Dropbox serve as the ultimate storage medium, unless we develop a different way of syncing files.