We did a lot of prototyping on AWS and still use them a lot, but really we are pretty agnostic as to where the VM lives. We need nvidia GRID cards (which is what we use to scrape the remote desktop really fast and which AWS has for prototyping) but the rest is pretty off the shelf. So currently its a pretty hybrid system
Not just a problem with developing countries. I have 100 Mb at home so that's not a problem, but the main use case I see for this for me is to able to use it while traveling. Even in countries with a great high speed internet in general, getting great high speed internet in a hotel room or conference center is almost impossible.
Hopefully it'll scale down gracefully when not doing graphics-intensive work. RDP is quite usable even with 100ms latency and only a few Mbps. OTOH I guess nothing's stopping you from renting a machine from Amazon now.
I'd just be worried if they use H.264 that there's not lossless compression on most UI elements. RDP falls back to lossy video-compression like behaviour under pressure, and it looks terrible and quite unusable. For Photoshop and graphics apps, sure. But for text and most UI elements, ouch.
The copy on your website ranges for me from confusing to offputting. I'm not even sure exactly what you're selling after having looked at your site for two minutes. Maybe you sell virtual machines in a data center, accessible with vnc or rdp implemented in JS? And maybe an additional mini computer preconfigured to access that?
Note: The answers to that do not matter. What matters is that your website doesn't answer them and is in fact so vague that it triggers all of my mental fraud bells.
Edit:
Also small footnote. This is probably honest and well-meant, but it reads like snake-oil: "Your desktop is managed in a secure data center and we send you a fully-encrypted feed." Which data center? What does secure mean? What guarantee do i have about that security? What encryption are you using? What about the data i put on my VM, are you backing it up? Is it encrypted by default or can any paperspace employee read my diary on my cloud computer?
You need to answer them in that sentence but without links and/or footnotes to further explanations it's just meaningless and amounts to "Just trust us, ok?"
Apologies for the confusion (and yes, the website needs some love). To answer the questions, we are primarily building a streaming protocol that makes remote desktops usable for a broader range of applications (namely "media-rich" use cases like photoshop, CAD/CAM, etc)
Behind the scenes we are using GPU tech originally developed for video game streaming, but using it primarily as a way to send a vanilla desktop.
We are currently delivering this desktop to any webbrowser using a JS renderer. That said, we didn't want you to have to have an old machine to use this one, so thats where the paperweight comes in.
VDI/remote desktops have been around for a while but they are usually really poor quality/hard to setup/etc. So we are trying to wrap that all in a simple package.
Thanks for clarifying. This is a much better explanation than the copy on the website. I could see this being useful for me as a developer, especially if you make it as easy to spin up an isolated virtual Windows/Linux desktop as it is to spin up a server on something like Digital Ocean.
I think you guys might be doing yourselves a bit of a disservice by promoting the hardware before you are in a position to really provide clear information about the actual service and the pricing. It creates a lot of confusion about what it is that you actually offer.
Especially considering the fact that the site recommends a minimum of >20mbps download speed... I suspect that anybody with that kind of connectivity already has a computer.
One more question: how are you handling the Windows licensing?
> Maybe you sell virtual machines in a data center, accessible with vnc or rdp implemented in JS? And maybe an additional mini computer preconfigured to access that?
Maybe it's just that I'm not interested in whether it's any of those things, but I have no problem with the copy on the website. The site says to me: we will sell you a computer, keep it at our place, and let you use it from yours. And it's secure.
Don't take this the wrong way but you might be coming at it from a too-technical, less consumer-based focus.
In the video you claim that it is "secure". I understand that as "nobody else can see my files or what I do, guaranteed".
Well, you could place a bug between the VNC and the VPS parts. Or your government might force you to do it. How can you reassure me that isn't the case?
We take privacy and security really seriously and have designed the brokering process (the part that connects you to your remote computer) to support this claim by following industry best practices.
We are still in a limited pilot program, but one of primary things we are testing is how to ensure that your computer is not compromised in any way.
We think that if we can achieve the technical goals then arguably, having a remote machine is more secure than one that someone can take from your house/car/etc.
That said, it is a really hard technical problem to manage a computer for someone without having any access to it and that is something we are designing out now.
That comment is really not reassuring.
I'm happy that you take privacy and security very seriously, that's great to hear.
However industry best practices is a little vague and 'to ensure that your computer is not compromised in any way' makes it sound like you want to do the job even the best antiviruses have trouble doing.
Furthermore I'd argue that if someone steals your machine from your house/car/etc, getting the keys to any remote machine should be trivial just by looking at the stored passwords and configuration.
But maybe I misunderstood what you were saying, sorry if that's the case.
'Industry Best Practice' in this case amounts to 'we will hand all of your data over to the government' if required.
Abdicating responsibility for personal security to ANY 3rd party makes one less secure, period.
Now, when the government comes knocking, instead of handing over some files or access records, you'll be able turn over the user entire computer. That's the exact opposite of security to me.
Do you have any information to share on how much it will cost to use the service (with/without a Paperweight)? Is there a monthly subscription option, or is it pay-as-you-go?
"...but support for other operating systems, including Mac OS X, are planned." – Techcrunch
How feasible do you think supporting OS X will be? Wouldn't Apple be hesitant (more than hesitant) to cooperate since hardware is their source of profit? Unless you are thinking about reverse engineering the Cocoa WindowServer?
Looks a lot like the old workspot [0] which I really loved (when it worked) so that's cool.
But why don't you lead with the Paperweight offering and just cut out all the other stuff? Just throw the lady's computer on the floor, and then give her the Paperweight. For me, that's the attractive bit. I could definitely sell that to clients because it implies 2FA security.
Let me guess, you're using webrtc to deliver a screencast of a headless VM to my browser plus JS to capture I/O events and deliver them over a websocket to the server essentially replicating the functionality of a "zero client" using modern web technologies, no?
Is the paperweight optional? I'd like buy the paperweight for my father, but then be able to access his paperspace from my computer when he needs help on something.
The paperweight is completely optional. In your case, it would make sense for him to use one and then you could access his desktop through your web browser. One of neat things about streaming desktops is that you get screen sharing built in
What do you get for the pre-order? The link takes me directly to a credit card page. Do you get a paperweight + 0 months of service? Or does it include a few months to get started?
This looks really interesting. I'd like to know more about network lag. Even with a better protocol, you can't get past the physical speed of a network. How laggy is paperspace in practice?
Thanks for stopping by to answer questions. My question is about privacy...
Since you must sign-in to use Paperspace (and everything you do is saved or performed online), do you track user activity in any way? For example, apps used, tasks performed, sites visited.
If you do track, what kind of data do you collect? Is the tracking data tied to individual accounts? Or disassociated from them? How long do you keep this data? Is it anonymised?
Which OS do you support? Some screenshots or demo videos of the workspace would be nice.
Edit: Also, a comment on the video - it started off well with the woman having issues with an old slow computer (which is what I have and so I could relate). But then later on all the actual use is shown on fancy macbooks and iMacs which kind of defeats the puropose for me.
Do you have an option to start up with a clean and fully patched system each time and not retain state? Can users have a library of machines for different purposes? I am thinking that something like this could be valuable as a secure and disposable space for someone living or traveling in a risky location.
Have you tested it on a phone with an external display?
I want something touchable about latency (I tried lots of remote working but it doesn't work "coz" latency). Something like, the usual remote networking app runs on 10 frames per second, we run on 60. Or, given the connection is the same, we are 10-20-50 times better/faster...
Yea, so VNC is really versatile but not very performant. We set out to build a remote desktop protocol that could stream HD video streams and which would make doing things like high-end graphics and other media-rich applications possible.
On a more technical level, VNC sends block of pixels whereas we are using a video stream.
Paperspace founder here. You can take snapshots at any time and we offer an instant rollback to a safe state in the event something happened. One of the neat benefits of virtual machines over traditional desktops.
As much I understand, the service (not the device) is not generally available?
When will it be ready? When will it be possible to use it from other countries (other than US)?
The service as such is very interesting, for example, when I want to have quick access to a Linux/Windows desktop with a particular OS without the need to install on real iron.
Depends of course on the pricing. The idea sounds really neat!
A Chromebook adjunct. A way to get Windows and its apps on the Chromebook. The lack (or difficulty) thereof is what stands between me and owning one.
I'm considering teaching, selling and supporting Chromebooks and the people using them in "old folks homes" (as we called them long, long ago) as an inexpensive and low skill entry to the internet. Some potential clients will have enough experience that they will require Windows access. If (and it is a bit hard to determine from the site) Windows machines with persistence can be created that problem is solved.
Have you looked at the issue of migration from an existing Windows home system into one of yours? Were that made relatively easy it would be a huge benefit. Perhaps a partnership with Laplink and their PCmover would benefit both you and them.
So, a thin client connecting to a virtual machine under some other company's control? No, thanks.
The fact is, computers are plenty fast now, and can handle 90% of peoples' usage patterns (browsing reddit, watching cat videos) with ease. The last thing I want is to have a remote desktop that lags out whenever there's a network hiccup.
I'm having a weird issue where my modem keeps restarting. I'll fix it, but it's not bad enough to worry about. With paperspace, I get to break into a cold sweat because I can't access my shit!
General computers are undoubtedly getting fast enough for a lot people. However, hosted computers (VDI, DaaS, etc) can do things that regular computers can't.
I hate having to upgrade a physical computer (which isn't even an option on laptops usually) and with a VM that is no longer a concern. If you need more power (which admittedly not everyone does) its really is a click away.
Transferring large files is also something that is still not easy to do with existing cloud storage options because the model is flipped (both parties have to upload and download over a single connection). By contrast, when using two hosted paperspace machines, you can transfer files nearly instantaneously. Effectively you trade one type of network transfer (the remote desktop image) for another (a big file you might be collaborating on with someone) and in some cases thats a huge win.
I very much like Browserstack for doing this as it allows for easy testing. I just don't like Browserstack's pricing as no shop I've worked in, and my own company, has ever needed it continuously. Rather, we need sorta pay-as-you-go minutes like cheap cell phones
This is not about the cost of hardware or software, it's about the price (and hassle) of maintaining them. Also, it's not about browsing reddit and accessing gmail; it's more about 'real' office applications and the like.
In certain non-tech-savvy environments, such a thing has a very good chance of adoption.
Also, it's a nice venue for renting heavyweight applications (like CAD, graphics, finance, etc) for a limited time, with a full guarantee against violating the policy by users.
And having to find out that some obscure feature you were using is now gone, and some other semi-obscure feature you were using is now in a different location and you have to rework your entire workflow yet again. It gets tiresome after twenty years.
I see people complaining about the hassles of maintaining a computer. Seriously? It's been like... how many years since basically everyone introduced that "revert to when this was working" state?
And even if it weren't for that, if maintaining a computer is so much of a hassle, how exactly is this supposed to be accessed? With browsers running on the holodeck? "From a consumer-level device, like a tablet"? Yes, I can't wait to use AutoCAD with a pointing device that has a precision of ughhh about my thumb's size.
And let's not get into the whole privacy thing. A 1 TB drive is, what, fifty bucks today -- and people would rather store their files in a datacenter across the ocean?
You’re thinking about how people are currently using their devices instead of thinking about how they will be. You could access your computer from any internet connected device, anywhere.
Hook up your oculus rift to their paperweight and have a full powered computer anywhere (with internet access).
While they are mentioning you can use AutoCad, I'd say that is merely to show that it can handle intensive applications. The majority wants to be able to surf the internet, and with this, you would never need to buy another computer again. For the random times you need to run Photoshop it can do it and will scale up to it, other times it scales down to handle your reddit addiction.
Look at how many people use Dropbox, iCloud, Google drive, etc etc. Everyone is already ok with storing their files in a datacenter. Privacy issues or not, people have accepted it is fact and the high majority already do it. That's not to say it is right, but it's a given that people want convenience over privacy.
Buying a Macbook air costs around 1k. And most people seem to upgrade ever 4-6 years. This costs $10/month. The cost savings are blatant. I know I’m comparing it with a mac and they are overpriced and you could do so much better, blah blah blah.
The fact is that if latency problems can be overcome, there is really no reason beside privacy that one needs to own and maintain their own computer. Rented space that you pay monthly for is cheaper, and will adapt to fit your specific needs. It works much better for a internet connected society to share several supercomputers then for individuals to each own their own.
If data provides competitive advantage, while compute is commoditized, there is no need for a false dichotomy between cloud and local. Future hybrid architecture can fluidly move the line between local and distributed compute, including other-peoples-algorithms, fusing local (secure, private, competitive, licensed) data/algos with public data/algos.
Local storage and local compute continue to fall in cost.
As long as humans remain sensitive to subsecond latency of repeated stimulus, and the speed of light remains unchanged, local user interfaces will offer competititve advantages.
> Future hybrid architecture can fluidly move the line between local and distributed compute, including other-peoples-algorithms, fusing local (secure, private, competitive, licensed) data/algos with public data/algos.
At the risk of sounding like a grumpy graybeard, it's impressive how all of these things have been implemented at least in Plan 9 for almost twenty years now.
Thanks for the pointer. Could you recommend any Plan9 docs on these features? Are Plan9 users mostly running on bare-metal or VMs? If the latter, would you recommend VMware, Xen or something else?
The Plan 9 documentation is pretty good, but 9front's introductory documents ( https://code.google.com/p/plan9front/wiki/fqa ) are probably a little better to get your feet wet with.
> Are Plan9 users mostly running on bare-metal or VMs? If the latter, would you recommend VMware, Xen or something else?
Both, I guess... VirtualBox is the only one I tried it on. Its distributed nature makes it fairly easy to run it on a server and connect to that from a system running Plan 9 from User Space (see http://swtch.com/plan9port/ ), so that you also get a... well, a functional desktop, I guess.
> You’re thinking about how people are currently using their devices instead of thinking about how they will be. You could access your computer from any internet connected device, anywhere.
No, I'm thinking about how people are going to be using their devices. One of the advantages this approach touts is that you can have a generic, cheap machine, and just rent something more powerful when you need it.
CAD is the obvious example (which they are giving, not me!) about why this can't really be done: you'll need a fast display and a lot of processing power just to run the browser in the first place (which, with Firefox and Chrome as reference, kind of discards anything older than 4 years anyway), and then you're going to need specific input devices, so it won't really be "any" computer.
Why do you think there were CAD workstations even in the 80s and early 90s, which were the bloody heyday of thin clients and distributed apps? Turns out that, when 80% of the price of the device is due to the graphics cards and monitors, slapping RAM and a hard drive on it makes a lot of things easier.
> Hook up your oculus rift to their paperweight and have a full powered computer anywhere (with internet access).
And what exactly is going to render things for the Oculus Rift? I'm pretty sure you need a computer for that as well!
> While they are mentioning you can use AutoCad, I'd say that is merely to show that it can handle intensive applications.
We have known that intensive applications can be used over a network for at least thirty years. I'm not surprised. What I haven't seen yet -- and this project does not seem to address -- is ways in which these applications can be low- (or at least constant-)latency or how remote storage can simultaneously be secure, private and permanently accessible. I've used networked applications over network a lot faster than the 15 Mbps that Paperspace recommends. Experiencing that over the Oculus Rift makes me nauseous just by thinking about it :).
> The majority wants to be able to surf the internet, and with this, you would never need to buy another computer again.
I doubt it. It's going to be the same 4-6 years before the old Paperspace One will stop being able to keep up with the higher-resolution, more bandwidth-demanding applications (even if they're executed remotely) and you'll have to upgrade to Paperspace Two. The "just wants to surf the Internet" majority would benefit a lot more from fixing web browsers.
> Look at how many people use Dropbox, iCloud, Google drive, etc etc. Everyone is already ok with storing their files in a datacenter. Privacy issues or not, people have accepted it is fact and the high majority already do it. That's not to say it is right, but it's a given that people want convenience over privacy.
I'm not questioning that -- and I'm not questioning the business model of Paperspace. I'm questioning the actual value that it delivers to users -- I'm not questioning the fact that they're willing to pay for it.
> The fact is that if latency problems can be overcome
I think we're optimistically 10 years away from that :). Given the circlejerk happening in telecom and software development circles (especially the slower ones), I think we're a good twenty years before that kind of latency becomes feasible for full-screen, full-color, multimedia-oriented applications under this execution model.
I disagree with the "Dropbox, iCloud, Google drive" thing. Nobody is using them as a platform to work their files on, but more as a quick way of sharing files. not reliable to store or else.
Definitely not "nobody". I've had several clients where the dropbox folder was where work got done. For all of my non-git and non-movie projects, dropbox is where the files live and are worked on.
When a friend had his laptop stolen he said that it wasn't a big deal other than the money lost, as 90% of his files were in dropbox or Google drive.
In my world at least, this is the way the wind is blowing.
I can access my computer (computers, actually, all at home) now, from any internet connected device, anywhere. My "laptop" is an iPad with a keyboard, and with that, I can do remotely what I can do locally. It's just as convenient [1], no recurring charge and if the government (or law enforcement) ever becomes interested in my data, I'll have a chance to know about it, since my data resides at my house, not some company half a continent away.
[1] Nearly so---for me, there's no true replacement for an IBM model M keyboard.
> And let's not get into the whole privacy thing. A 1 TB drive is, what, fifty bucks today -- and people would rather store their files in a datacenter across the ocean?
People would rather have convenience of access across their devices. Despite the fact that people tend to have home networks and go around with a wifi-enabled device in their pocket, syncing tools remain atrocious.
That's the pain that dropbox addresses, not the cost of storage.
> And let's not get into the whole privacy thing. A 1 TB drive is, what, fifty bucks today -- and people would rather store their files in a datacenter across the ocean?
Yep. Fact is, HNers are much more concerned with privacy than your average consumer is. Most people care much more about convenience than privacy. I don't really know if that's a good or a bad thing, but it's a truth.
I can easily create a Windows or Linux desktop machine in the cloud at Digital Ocean or Vultr.com for very little money and access it with rdesktop and/or a VPN. Amazon has a slightly more "enterprise" desktop-in-cloud offering that's basically the same idea.
A lot of the underlying technology for paperspace certainly exists today and its best to think of what we are doing as providing a layer on top of it all (and polishing the pieces that didn't work so great before):
To give an analogy, cloud storage was definitely around long before dropbox and google drive. It was simple, I uploaded a file to FTP, you opened your FTP client, and we could share files. The reality was that less tech-savvy people might never get past the install step.
What dropbox/google drive/box did was take an existing (and increasingly commodity) offering and wrapped it in a layer that made it accessible to a whole new class of users.
Digital Ocean is amazing and we are shooting for the same simplicity for full desktop computers.
yo dawg I heard you need a computer to use your computer. Jokes aside, tablets/phones are carried with us everywhere and for on the go can accomplish most of what one may do when they are away from a dedicated pc. I can ssh into any one of my computers and do whatever is needed from my phone for the most part.
I'm really curious as to what the main use case would be ? I think this would make big strides in saving licensing costs if it were possible for certain cases where a shared computer may make sense.
I'm sorry, is this a running joke on HN? The first reply is always some bland, obvious attack on whatever is posted. I mean, the fact that a streaming computer in the cloud is sensitive to network hiccups is so obvious to anyone who sees this it's not even a valid criticism.
Actually, I wasn't the first reply. TBH I was expecting this to be buried since I was so late to the party.
But regardless, my criticism is that slow speed in the hardware is rarely a problem these days. Almost every "slow computer" problem I've diagnosed in recent times (that hasn't been due to McAffee) has in fact just been a slow internet connection or a misconfigured router. When you're not a computer expert, it's hard to know why your Youtube video stutters.
So yes, they will probably get a lot of sign ups with a mantra of "faster", but IMO this isn't solving the actual problem that users tend to have with computer "slowness". In fact, it will only exacerbate it, although that might be good for pushing towards greater bandwidth.
I'm also not too keen on putting even more of my data on someone else's system.
What if I muck up my space, can I go back to a fresh install? Which flavor of linux?
Where are my files stored? Say I want to take those files on the plane with me and work from my laptop, will I need something like Dropbox/OneDrive/SpiderOak?
Paperspace founder here. We offer an instant rollback to a safe state which is a neat thing that VMs can do that regular computers can't (or at least don't do very well). Everything is stored in the cloud so have your entire desktop available to you at all times meaning you wouldn't need to transfer your stuff to a cloud storage service. It's actually just like cloud storage but with the added benefit of having your whole operating system (apps, settings, shortcuts) on the go.
Somewhat off-topic, but it's kind of disappointing that the intro video doesn't play on my copy of Firefox nightly on Linux -- Kickstarter videos play fine, since it seems like they provide a webm video.
Error says:
Awww, snap!
This video can’t be played with your current setup.
I could open up Chrome or Opera to play the video, but that's too much friction for something I might not care too much about.
It _looks_ like Vimeo -- I wonder if they let you know how many people tried to view the video but couldn't.
This seems like a complete homerun for small to medium sized businesses like the one I'm working in now, provided that's a major focus and their specific needs are met (monitoring, backup, security).
Absolutely. If you are a very large company (lets say >500 people) then you can buy some expensive and complicated tools from the big guys to handle this kind of thing. We are trying to wrap the hard parts of administering desktops into a really easy interface so that i.e. redundant and regula backups are no longer a big worry
Cool! I've worked with similar technologies in the past. Most common issues are with peripherals and web browsing (sticky frames) in enterprise environments. Ncomputing came closest to what I was looking for. Integrates perfectly with printers, scanners, barcode scanners, usb sticks, audio etc. Small-form factor with wiFi (not all models) and you are able to host the instances right in your DC or LAN.
I've done A LOT of research and experimentation in this space over the course of the last 5-ish years, mainly out of sheer curiosity. This sounds very, very similar to something I prototyped a while back using x264 + Broadway.js (HTML5 video streaming with low latency is a non-starter).
I'm extremely curious, has Paperspace actually developed an entirely new video codec, encoder, and efficient JS decoder for this?
I'm a little confused by the statements so far (and the comments are flowing in, so apologies if this was addressed while I was typing): "we are using a video stream", "using a JS renderer", "building a streaming protocol", "using GPU tech originally developed for video game streaming", "a remote desktop protocol that could stream HD video"
So, is just the streaming protocol you've developed and not the codec? In my expiriments, simply pushing out h264 NAL units over websockets and passing them to the decoder was a pretty solid start. Add a tiny layer of buffering over that and I imagine it'd be fairly stable. Ultimately, I backed away from h264 for licensing and performance issues.
Also, what's the transport into the browser? Websockets? WebRTC data channels? Have you encountered performance issues with Firefox not being able to handle websockets or data channels in web workers? (Which is seemingly coming in FF37.)
awesome questions! There will be tech blog deep dive that goes into more detail, but just to clarify things a bit we are using h264 as the underlying codec but the protocol is a combination things -- h264 packets over websockets is where we started too!
Usually you want a "reliable" transport (tcp) for handshake, some transfer, etc and then you can get away with a more fire-and-forget (udp) stream for everything else. We broker different connections for different scenarios. The web version is naturally limited to either webrtc or websockets but for the paperweight it didn't make sense to force a web paradigm (or full webbrowser just to access limited socket types) so we interface directly with the hardware.
Haven't encountered big problems with firefox, but browser inconsistencies are definitely something we spend a lot of time working through
Because Paperweight is the black circle hardware device they made to stream from Paperspace. There's no point to using a browser on it, just stream directly through the wifi/ethernet port using TCP/IP.
I'm curious what performance issues you were seeing in Firefox. I'm doing quite a bit of work with the same stack, and have been able to get sub-10ms decode times with Broadway.JS in all the major browsers. As long as you use transferable objects to pass data around, I haven't had any problems using web workers as decode threads either.
I'm sorry, I actually mixed up two different experiments regarding the performance issues (I've been playing with this stuff for a looongg time and have many different prototypes). Allow my to clarify. :)
The performance issues with using Broadway.js were more "general", in that the experience varied wildly, from amazing to unusable depending on the browser/device/etc)... Not necessarily related to Websockets + Webworkers.
The Websocket issue on the other hand, was for a more recent experiment where I was playing with the idea of bridging NoVNC over WebRTC data channels.
It's not so much a performance "issue" as much as a "possible area for optimization"... Though, every time I'm playing with noVNC or Broadway.js in a FF tab on my i7 laptop, it pretty much renders FF pretty laggy in all other tabs. I imagine offloading as much of the processing to workers as possible would be the best approach to lessen the effect — though, I'm not much of a frontend / JS dev.
Shameless plug: check out jsmpeg[1] - it's an MPEG1 decoder written in JavaScript that's capable of low latency streaming via WebSockets, weighs only ~25kb gzipped, offloads part of the decoding to the GPU via WebGL and runs smoothly on an iPhone4.
Of course it has some drawbacks, as MPEG1 is a pretty old codec. The data rate is fairly high and it struggles with high resolution streams. 720p still decodes nicely in realtime on most mobile devices, though.
Thanks for sharing! I've definitely bumped into jsmpeg a number of times when researching JS decoders. Did you ever pursue an emscripten/ asm.js version?
Is this designed to compete with thin client offerings from the big boys when it comes to larger installs?
One big advantage for something like microsoft remote desktop services is ability to throw a beefy server into an existing infra with aging desktops and use them as thin clients/dumb terminals.
I can see an advantage of this as being able to take just that box when i travel, plug it into any random machine, and get an instant resume of my previous desktop session over VPN... but this makes it oh-so-close to Windows To Go?
Most importantly, what is the monthly cost of the actual VPC (not the hardware)? Its not listed anywhere on the site that i could find. What would you do better than Azure, Amazon VPC, rackspace, or multiple others?
Does it say what the computer is full with? I would probably be not that interested if it is bees. In contrast, if it is dark chocolate, I would be very interested.
I love this idea. I'm very tempted to pre-order but the lack of a pricing page keeps away from pulling my wallet out.
Here is one use I'm really interested in. I'd love to get this for my father who is not very computer literate and needs a lot of help with it. He doesn't have strong security concerns because he only uses it for email and basic card games. Unfortunately I live hundreds of miles away and have a difficult time doing remote support for him. With this device+service it cleans up the clutter on his desk, and I hope allows for us both to connect to the VPS even though I don't have the paperweight. This way I can help him when he needs it and I can connect with ease.
I don't think doing that is pretty hard with today's technologies. I do the same to my father using a free version of TeamViewer and setting unattended access on his machine. I can access his machine whenever I want, fix things and close the session. I don't need anything more than that.
This would be useful in places with notoriously bad internet connections. When I was in China a couple years ago, even getting a reliable connection was a stretch, let alone being able to use it for something other than transfer of small plaintext.
I set up a box back at home and was able to use my 60MB connection by just transferring I/O via something like VNC or LogMeIn. I can see this being valuable for users in countries with subpar internet connectivity.
Thanks for the typo catch... Great use-case btw, it's surprising how well virtual desktops can perform on slow connections. I'm from the east coast but work in Mountain View now and I stream my desktop from Virginia everyday for work. Right now in fact haha.
213 comments
[ 2.1 ms ] story [ 195 ms ] threadFounder here. We just announced Paperspace today and I'd be happy to answer any questions (technical or otherwise).
P.S I am from India and its difficult to find more than 2 Mbps internet bandwidth with ease here.
Sad even I will not be able to use it.
I'd just be worried if they use H.264 that there's not lossless compression on most UI elements. RDP falls back to lossy video-compression like behaviour under pressure, and it looks terrible and quite unusable. For Photoshop and graphics apps, sure. But for text and most UI elements, ouch.
Funky way to denote minimum - reminds me of the "Recommended" specs on a videogame.
That rules out me, but it's a nice reminder of how much of an obstacle to innovation lack of broadband can be.
Note: The answers to that do not matter. What matters is that your website doesn't answer them and is in fact so vague that it triggers all of my mental fraud bells.
Edit:
Also small footnote. This is probably honest and well-meant, but it reads like snake-oil: "Your desktop is managed in a secure data center and we send you a fully-encrypted feed." Which data center? What does secure mean? What guarantee do i have about that security? What encryption are you using? What about the data i put on my VM, are you backing it up? Is it encrypted by default or can any paperspace employee read my diary on my cloud computer?
You need to answer them in that sentence but without links and/or footnotes to further explanations it's just meaningless and amounts to "Just trust us, ok?"
Behind the scenes we are using GPU tech originally developed for video game streaming, but using it primarily as a way to send a vanilla desktop.
We are currently delivering this desktop to any webbrowser using a JS renderer. That said, we didn't want you to have to have an old machine to use this one, so thats where the paperweight comes in.
VDI/remote desktops have been around for a while but they are usually really poor quality/hard to setup/etc. So we are trying to wrap that all in a simple package.
That makes me really look forward to when i can use that technology to remotely access my own machines from my cellphone. :D
I think you guys might be doing yourselves a bit of a disservice by promoting the hardware before you are in a position to really provide clear information about the actual service and the pricing. It creates a lot of confusion about what it is that you actually offer.
Especially considering the fact that the site recommends a minimum of >20mbps download speed... I suspect that anybody with that kind of connectivity already has a computer.
One more question: how are you handling the Windows licensing?
Maybe it's just that I'm not interested in whether it's any of those things, but I have no problem with the copy on the website. The site says to me: we will sell you a computer, keep it at our place, and let you use it from yours. And it's secure.
Don't take this the wrong way but you might be coming at it from a too-technical, less consumer-based focus.
Well, you could place a bug between the VNC and the VPS parts. Or your government might force you to do it. How can you reassure me that isn't the case?
We are still in a limited pilot program, but one of primary things we are testing is how to ensure that your computer is not compromised in any way.
We think that if we can achieve the technical goals then arguably, having a remote machine is more secure than one that someone can take from your house/car/etc.
That said, it is a really hard technical problem to manage a computer for someone without having any access to it and that is something we are designing out now.
Furthermore I'd argue that if someone steals your machine from your house/car/etc, getting the keys to any remote machine should be trivial just by looking at the stored passwords and configuration.
But maybe I misunderstood what you were saying, sorry if that's the case.
Abdicating responsibility for personal security to ANY 3rd party makes one less secure, period.
Now, when the government comes knocking, instead of handing over some files or access records, you'll be able turn over the user entire computer. That's the exact opposite of security to me.
But why don't you lead with the Paperweight offering and just cut out all the other stuff? Just throw the lady's computer on the floor, and then give her the Paperweight. For me, that's the attractive bit. I could definitely sell that to clients because it implies 2FA security.
[0] http://en.wikipedia.org/wiki/Workspot
Since you must sign-in to use Paperspace (and everything you do is saved or performed online), do you track user activity in any way? For example, apps used, tasks performed, sites visited.
If you do track, what kind of data do you collect? Is the tracking data tied to individual accounts? Or disassociated from them? How long do you keep this data? Is it anonymised?
Edit: Also, a comment on the video - it started off well with the woman having issues with an old slow computer (which is what I have and so I could relate). But then later on all the actual use is shown on fancy macbooks and iMacs which kind of defeats the puropose for me.
How can I?
Have you tested it on a phone with an external display?
Edit: additional question below...
Can I start long-running jobs (an image render) on it and then disconnect?
This is the largest thing that could happen (since it seems you are targeting people on-the-go from hotel to hotel).
I can't even begin to imagine the uses that this will have.
On a more technical level, VNC sends block of pixels whereas we are using a video stream.
We’re starting off with Windows and Linux options for now but plan to offer more operating systems in the future.
Anyways, the intro video is easily one of the more amusing ones I've seen in a long time. Props.
I can feel it in my head, it's like I'm "in battle" in a video game or something.
When will it be ready? When will it be possible to use it from other countries (other than US)?
The service as such is very interesting, for example, when I want to have quick access to a Linux/Windows desktop with a particular OS without the need to install on real iron.
Depends of course on the pricing. The idea sounds really neat!
I'm considering teaching, selling and supporting Chromebooks and the people using them in "old folks homes" (as we called them long, long ago) as an inexpensive and low skill entry to the internet. Some potential clients will have enough experience that they will require Windows access. If (and it is a bit hard to determine from the site) Windows machines with persistence can be created that problem is solved.
Have you looked at the issue of migration from an existing Windows home system into one of yours? Were that made relatively easy it would be a huge benefit. Perhaps a partnership with Laplink and their PCmover would benefit both you and them.
The fact is, computers are plenty fast now, and can handle 90% of peoples' usage patterns (browsing reddit, watching cat videos) with ease. The last thing I want is to have a remote desktop that lags out whenever there's a network hiccup.
I hate having to upgrade a physical computer (which isn't even an option on laptops usually) and with a VM that is no longer a concern. If you need more power (which admittedly not everyone does) its really is a click away.
Transferring large files is also something that is still not easy to do with existing cloud storage options because the model is flipped (both parties have to upload and download over a single connection). By contrast, when using two hosted paperspace machines, you can transfer files nearly instantaneously. Effectively you trade one type of network transfer (the remote desktop image) for another (a big file you might be collaborating on with someone) and in some cases thats a huge win.
In certain non-tech-savvy environments, such a thing has a very good chance of adoption.
Also, it's a nice venue for renting heavyweight applications (like CAD, graphics, finance, etc) for a limited time, with a full guarantee against violating the policy by users.
I see people complaining about the hassles of maintaining a computer. Seriously? It's been like... how many years since basically everyone introduced that "revert to when this was working" state?
And even if it weren't for that, if maintaining a computer is so much of a hassle, how exactly is this supposed to be accessed? With browsers running on the holodeck? "From a consumer-level device, like a tablet"? Yes, I can't wait to use AutoCAD with a pointing device that has a precision of ughhh about my thumb's size.
And let's not get into the whole privacy thing. A 1 TB drive is, what, fifty bucks today -- and people would rather store their files in a datacenter across the ocean?
Hook up your oculus rift to their paperweight and have a full powered computer anywhere (with internet access).
While they are mentioning you can use AutoCad, I'd say that is merely to show that it can handle intensive applications. The majority wants to be able to surf the internet, and with this, you would never need to buy another computer again. For the random times you need to run Photoshop it can do it and will scale up to it, other times it scales down to handle your reddit addiction.
Look at how many people use Dropbox, iCloud, Google drive, etc etc. Everyone is already ok with storing their files in a datacenter. Privacy issues or not, people have accepted it is fact and the high majority already do it. That's not to say it is right, but it's a given that people want convenience over privacy.
Buying a Macbook air costs around 1k. And most people seem to upgrade ever 4-6 years. This costs $10/month. The cost savings are blatant. I know I’m comparing it with a mac and they are overpriced and you could do so much better, blah blah blah.
The fact is that if latency problems can be overcome, there is really no reason beside privacy that one needs to own and maintain their own computer. Rented space that you pay monthly for is cheaper, and will adapt to fit your specific needs. It works much better for a internet connected society to share several supercomputers then for individuals to each own their own.
Local storage and local compute continue to fall in cost.
As long as humans remain sensitive to subsecond latency of repeated stimulus, and the speed of light remains unchanged, local user interfaces will offer competititve advantages.
At the risk of sounding like a grumpy graybeard, it's impressive how all of these things have been implemented at least in Plan 9 for almost twenty years now.
The Plan 9 documentation is pretty good, but 9front's introductory documents ( https://code.google.com/p/plan9front/wiki/fqa ) are probably a little better to get your feet wet with.
> Are Plan9 users mostly running on bare-metal or VMs? If the latter, would you recommend VMware, Xen or something else?
Both, I guess... VirtualBox is the only one I tried it on. Its distributed nature makes it fairly easy to run it on a server and connect to that from a system running Plan 9 from User Space (see http://swtch.com/plan9port/ ), so that you also get a... well, a functional desktop, I guess.
No, I'm thinking about how people are going to be using their devices. One of the advantages this approach touts is that you can have a generic, cheap machine, and just rent something more powerful when you need it.
CAD is the obvious example (which they are giving, not me!) about why this can't really be done: you'll need a fast display and a lot of processing power just to run the browser in the first place (which, with Firefox and Chrome as reference, kind of discards anything older than 4 years anyway), and then you're going to need specific input devices, so it won't really be "any" computer.
Why do you think there were CAD workstations even in the 80s and early 90s, which were the bloody heyday of thin clients and distributed apps? Turns out that, when 80% of the price of the device is due to the graphics cards and monitors, slapping RAM and a hard drive on it makes a lot of things easier.
> Hook up your oculus rift to their paperweight and have a full powered computer anywhere (with internet access).
And what exactly is going to render things for the Oculus Rift? I'm pretty sure you need a computer for that as well!
> While they are mentioning you can use AutoCad, I'd say that is merely to show that it can handle intensive applications.
We have known that intensive applications can be used over a network for at least thirty years. I'm not surprised. What I haven't seen yet -- and this project does not seem to address -- is ways in which these applications can be low- (or at least constant-)latency or how remote storage can simultaneously be secure, private and permanently accessible. I've used networked applications over network a lot faster than the 15 Mbps that Paperspace recommends. Experiencing that over the Oculus Rift makes me nauseous just by thinking about it :).
> The majority wants to be able to surf the internet, and with this, you would never need to buy another computer again.
I doubt it. It's going to be the same 4-6 years before the old Paperspace One will stop being able to keep up with the higher-resolution, more bandwidth-demanding applications (even if they're executed remotely) and you'll have to upgrade to Paperspace Two. The "just wants to surf the Internet" majority would benefit a lot more from fixing web browsers.
> Look at how many people use Dropbox, iCloud, Google drive, etc etc. Everyone is already ok with storing their files in a datacenter. Privacy issues or not, people have accepted it is fact and the high majority already do it. That's not to say it is right, but it's a given that people want convenience over privacy.
I'm not questioning that -- and I'm not questioning the business model of Paperspace. I'm questioning the actual value that it delivers to users -- I'm not questioning the fact that they're willing to pay for it.
> The fact is that if latency problems can be overcome
I think we're optimistically 10 years away from that :). Given the circlejerk happening in telecom and software development circles (especially the slower ones), I think we're a good twenty years before that kind of latency becomes feasible for full-screen, full-color, multimedia-oriented applications under this execution model.
When a friend had his laptop stolen he said that it wasn't a big deal other than the money lost, as 90% of his files were in dropbox or Google drive.
In my world at least, this is the way the wind is blowing.
I've seen stuff covered under bloody NDAs being carelessly shoved into Dropbox or Google drive.
No one serious about their work is using them as a platform to work their files on, but they are used for that.
[1] Nearly so---for me, there's no true replacement for an IBM model M keyboard.
People would rather have convenience of access across their devices. Despite the fact that people tend to have home networks and go around with a wifi-enabled device in their pocket, syncing tools remain atrocious.
That's the pain that dropbox addresses, not the cost of storage.
Yep. Fact is, HNers are much more concerned with privacy than your average consumer is. Most people care much more about convenience than privacy. I don't really know if that's a good or a bad thing, but it's a truth.
To give an analogy, cloud storage was definitely around long before dropbox and google drive. It was simple, I uploaded a file to FTP, you opened your FTP client, and we could share files. The reality was that less tech-savvy people might never get past the install step.
What dropbox/google drive/box did was take an existing (and increasingly commodity) offering and wrapped it in a layer that made it accessible to a whole new class of users.
Digital Ocean is amazing and we are shooting for the same simplicity for full desktop computers.
I'm really curious as to what the main use case would be ? I think this would make big strides in saving licensing costs if it were possible for certain cases where a shared computer may make sense.
https://news.ycombinator.com/item?id=5072224
But regardless, my criticism is that slow speed in the hardware is rarely a problem these days. Almost every "slow computer" problem I've diagnosed in recent times (that hasn't been due to McAffee) has in fact just been a slow internet connection or a misconfigured router. When you're not a computer expert, it's hard to know why your Youtube video stutters.
So yes, they will probably get a lot of sign ups with a mantra of "faster", but IMO this isn't solving the actual problem that users tend to have with computer "slowness". In fact, it will only exacerbate it, although that might be good for pushing towards greater bandwidth.
I'm also not too keen on putting even more of my data on someone else's system.
What if I muck up my space, can I go back to a fresh install? Which flavor of linux?
Where are my files stored? Say I want to take those files on the plane with me and work from my laptop, will I need something like Dropbox/OneDrive/SpiderOak?
Liked the video as well.
Error says:
Awww, snap!
This video can’t be played with your current setup.
I could open up Chrome or Opera to play the video, but that's too much friction for something I might not care too much about.
It _looks_ like Vimeo -- I wonder if they let you know how many people tried to view the video but couldn't.
I'm extremely curious, has Paperspace actually developed an entirely new video codec, encoder, and efficient JS decoder for this?
I'm a little confused by the statements so far (and the comments are flowing in, so apologies if this was addressed while I was typing): "we are using a video stream", "using a JS renderer", "building a streaming protocol", "using GPU tech originally developed for video game streaming", "a remote desktop protocol that could stream HD video"
So, is just the streaming protocol you've developed and not the codec? In my expiriments, simply pushing out h264 NAL units over websockets and passing them to the decoder was a pretty solid start. Add a tiny layer of buffering over that and I imagine it'd be fairly stable. Ultimately, I backed away from h264 for licensing and performance issues.
Also, what's the transport into the browser? Websockets? WebRTC data channels? Have you encountered performance issues with Firefox not being able to handle websockets or data channels in web workers? (Which is seemingly coming in FF37.)
Usually you want a "reliable" transport (tcp) for handshake, some transfer, etc and then you can get away with a more fire-and-forget (udp) stream for everything else. We broker different connections for different scenarios. The web version is naturally limited to either webrtc or websockets but for the paperweight it didn't make sense to force a web paradigm (or full webbrowser just to access limited socket types) so we interface directly with the hardware.
Haven't encountered big problems with firefox, but browser inconsistencies are definitely something we spend a lot of time working through
The performance issues with using Broadway.js were more "general", in that the experience varied wildly, from amazing to unusable depending on the browser/device/etc)... Not necessarily related to Websockets + Webworkers.
The Websocket issue on the other hand, was for a more recent experiment where I was playing with the idea of bridging NoVNC over WebRTC data channels.
It's not so much a performance "issue" as much as a "possible area for optimization"... Though, every time I'm playing with noVNC or Broadway.js in a FF tab on my i7 laptop, it pretty much renders FF pretty laggy in all other tabs. I imagine offloading as much of the processing to workers as possible would be the best approach to lessen the effect — though, I'm not much of a frontend / JS dev.
Here's the noVNC issue: https://github.com/kanaka/noVNC/issues/114 And the bugzilla for FF: https://bugzilla.mozilla.org/show_bug.cgi?id=504553
Of course it has some drawbacks, as MPEG1 is a pretty old codec. The data rate is fairly high and it struggles with high resolution streams. 720p still decodes nicely in realtime on most mobile devices, though.
[1] https://github.com/phoboslab/jsmpeg
Is this designed to compete with thin client offerings from the big boys when it comes to larger installs?
One big advantage for something like microsoft remote desktop services is ability to throw a beefy server into an existing infra with aging desktops and use them as thin clients/dumb terminals.
I can see an advantage of this as being able to take just that box when i travel, plug it into any random machine, and get an instant resume of my previous desktop session over VPN... but this makes it oh-so-close to Windows To Go?
Most importantly, what is the monthly cost of the actual VPC (not the hardware)? Its not listed anywhere on the site that i could find. What would you do better than Azure, Amazon VPC, rackspace, or multiple others?
Here is one use I'm really interested in. I'd love to get this for my father who is not very computer literate and needs a lot of help with it. He doesn't have strong security concerns because he only uses it for email and basic card games. Unfortunately I live hundreds of miles away and have a difficult time doing remote support for him. With this device+service it cleans up the clutter on his desk, and I hope allows for us both to connect to the VPS even though I don't have the paperweight. This way I can help him when he needs it and I can connect with ease.
Even your line "using GPU tech originally developed for video game streaming" makes me think you've actually licensed or are a subsidiary of OnLive.
Paperspace replaces hight upfront infrastructure costs with low monthly payments.
I set up a box back at home and was able to use my 60MB connection by just transferring I/O via something like VNC or LogMeIn. I can see this being valuable for users in countries with subpar internet connectivity.