Amazon tries again with AppStream because customers didn't like it Acknowledging that the initial version of application streaming service AppStream failed to appeal to customers, Amazon Web Services is ready to try again. Introduced in 2013, AppStream offered companies a way to stream Windows desktop apps to users via browser across a variety of devices. It was intended as an alternative to …
This idea recirculates every couple of years. It used to be the mainframe and 'dumb' terminals. Then it was "thin" clients. Then it was Citrix. Zenworks. This has been tried over and over again, and it always flops.
It all comes down to latency, and to a lesser extent, bandwidth. The delay from when you take an action in a virtual environment until that action registers is perceptible to people; Even as little as 30 milliseconds, a fairly typical latency on the internet, is noticable... and people conclude that it is slower than their own computer. And it's a giant bag of "Nope" then. It never really catches on because while the computer can cope with lag, the human brain? Not so much.
But even if we used magical pixie dust to erase that problem, people have been trying to shove everything into the browser pretty much since the Netscape 2.0 days. That idea also digs itself out of the grave every couple of years and shambles across the landscape, and the fact is... the browser is a terrible platform to develop complex interactive applications on. Microsoft has even tried streaming XBox games to the desktop... and even then, when it's literally in the same room... not many people write rave reviews about it. It may simply be that this is a human perception issue -- even with all the problems of latency, bandwidth, etc., solved... people still just don't seem to want it.
"AppStream" died before it ever launched, or re-launched... and it'll die again on its re-re-re-launch, because it's a solution in search of a problem.
> Yep, but it's different this time. Nothing much new has arrived in desktop software for some time so upgrade revenues have collapsed. All tech companies are now desperately casting around for something with recurring revenues rather than coming up with genuine new ideas.
It collapsed because there is no way to innovate when every possible invention and permutation of inventions have 150+ year copyright and patents attached to them. So now they're taking what we already have, and moving them to a subscription-based economy, and our idiot legislators are content to let them do it because "what's good for business is good for the country". The internet, and IT in general, is rapidly becoming the "Information Super Tollway".
"The four most dangerous words in investing are: 'this time it's different.'" - Sir John Templeton
Templeton's statement applies to a lot more things than investing, and this is one of them.
The thing that will finally push the world away from Windows is when Microsoft switches the OS for ordinary users to the subscription model, which will happen sooner rather than later.
Well, thin clients, either using terminals, Browser or an Application or entire OS (ChromeOS) is a niche. It's valuable to some people with decent broadband and one or two screens with no IT expertise in geographically dispersed offices.
I don't see a case for using it off Amazon unless you are already using their cloud to a foolish extent for core applications. Using Cloud to deliver Web sites and eCommerce to Retail consumers makes some sense, though Co-lo is "better" and maybe cloud for sales peaks?
Frankly, but for occasional use, I can't find a reason why this solution should be appealing to actual users.
For most of those tasks you still need expensive local hardware, so why run things remotely? Take professional image editing. You still need a specific monitor, an adequate video card - and tools to profile them, and probably you're also using specific input devices - tablets, Tangent consoles, etc. You'll also have specific and expensive local output devices, at least for proofing, if not for the final product.
Why should you run applications remotely on hardware you little control, adding a layer (the browser) which could add input/output issues - i.e. bitmaps are rendered by the server GPU and then re-displayed locally, looks to me a recipe for a lot of headaches trying to properly color manage that stuff. What about using specific input tools the browser doesn't know about? It would require a lot of client code just for it.
And all that on top of Internet latency and bandwidth.
While Citrix, Remote Desktop and VNC are not browser based, could be run from your switched gigabit LAN and not across the Internet through a single or few connections, still they are not the best environments for some applications. Sure some can be run without much issues, others can't.
The Amazon problems were clear:
""We thought game developers and graphics ISVs would embrace this development model" and ""Amazon suggests AppStream 2.0 can be used for CAD, 3D graphics, simulation, gaming, media editing, medical imaging, and life sciences applications.""
Wrong set of applications, no surprise nobody jumped on the wagon. While it can help to counter piracy, these are exactly some of the software which would have issues running within a remote view shown by a browser, because of latency (Amazon admits it too for games, but they are not the only sensitive ones), and because of non marginal issues like those I pointed out in an earlier post.
3D batch rendering? OK, but interactive modelling? Could it print to my expensive photo or 3D printer locally, for which it may need specific drivers and profiles?
Medical imaging? Do they understand what it means if an outage or connectivity issue cut doctors away from needed data? Once after surgery a doctor asked for my mother images - for some reason they weren't available on the LAN. Luckily I had the DVDs at hand, and the doctor could display them on his PC....
Sometimes the single workstation price is only a part of the hardware costs for such businesses (and for a smaller number of users), unlike most general business application where it could be the main hardware costs (and for a larger number of users).
Moreover ISV like Adobe would run their software from their own cloud, not the Amazon one, whenever they feel the need of a new cage for users.
" lots and lots of enterprises use citrix to push out a whole range of applications and have done for over a decade."
Yep. I like Citrix despite their incredible inability to make good admin interfaces.
It's particularly good for legacy apps, BYOD (/BYODT*) and some counter-intuitive uses like remote access to CAD products.
When it's the right sort of tool to be using nothing beats it - and I suspect that will be the problem for Amazon's product; the people who need it enough to pay for it will pay for something else.
*BYODT - Bring Your Own Dumb Terminal
" It never really catches on because while the computer can cope with lag, the human brain?"
You really should give the demo a try. Looks to me like they are using VMware's Blast protocol and even with 54ms latency (they give you an info widget to see bandwidth and latency) it's still really not bad. That have Visio, Excel, Eclipse, Firefox and a few other apps in the sample catalog. They launch almost instantly, feel very much like local apps, you can switch around open apps just like a real desktop, and overall it really wasn't a bad experience. I was fully prepared to be disappointed, and ended up thinking it's a viable solution for very specific use cases. The issue's going to be the same as every other AWS service - you pay for what you provision and not what you use, and this is no different. Gets very expensive very fast if you need it available all the time.
... because whatever the cost of this service I STILL NEED A PC, so how does this do anything except move the company's licensing costs on the spreadsheet, the IT dept still needs to provide about 90% of whats required.
And anyone who has been in office where there is construction near by will know that even in big cities a digger (or back haul ?) can break fibre without blinking.
Oh and now i have a quad core chip in my cheapy android phone and the next laptop will likely come with 8gb of ram minimum.... when will I be without my own computing power to run these apps?
AppStream 1.0 launched in late 2013, at the same time as Frame (www.fra.me). Where AppStream 1.0 admittedly failed, Frame successfully won over the major ISVs in CAD, digital content, and engineering, as well as many enterprise customers.
Frame's focus on enterprise and key early architectural choices (like being browser-based from day 1), allowed it to consistently beat AppStream, while giving customers the choice of running on any cloud, including AWS and Microsoft Azure.
So today, AppStream 2.0 freely borrows from Frame by copying its feature set and architectural choices from 3 years ago -- but the fact is that it remains years behind and not even close to the depth and maturity of the Frame platform.
Remember, cloud computing = somebody else's PC
What will happen during internet outages? Workers gonna sit around doing nothing while you have to pay their salaries/wages.
The same personnel will be still productive should you have the apps loaded on their PC's and the internet's gone to pot. I'll rather have this option than cloud "somebody else's PC" computing.
Because Mr Murphy.
Also, you don't need to have IT personnel 24x7 - you can always outsource if you really want to cut costs - but that comes with its own caveats. You will still need IT personnel even though you've gone with cloud computing.
Always.
There is no escaping that fact.
Because Mr Murphy. Yup. He's real.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
