Why Cant We Compute in the Cloud? - Bits - Technology - New Y...

Now, coupled with the rise of Google-style Web-based computing and the near-ubiquity of wireless broadband the time seems ripe for a new kind of computing.

Highlighted by eyalnow

Dozens of start-ups are creating an increasingly rich set of Web-based applications and more than a dozen efforts are under way to move the computer “operating system” itself onto the Web and away from the desktop.

Highlighted by eyalnow

For all the activity, however, one thing seems to be inexplicably missing.

There have been almost no credible efforts to design stripped down mobile computer hardware to match the wealth of Web software.

Highlighted by eyalnow

That said, nobody seems to be ready to really gamble on computing on the Web.

Highlighted by eyalnow

There is no privacy. Period.

Highlighted by eyalnow

I, for one, like the idea that my most important applications reside at a distance from the communal well…

Highlighted by eyalnow

I think there is a greater transition occurring to online computing than you’ve recognized. On-line computing for everything from basic word processing to file downloading and web publishing has been quietly taking off for years. Google’s free services alone let users save and edit text and spreadsheet documents, design and publish websites, maintain online photo albums, maintain calendars–all of which were, until somewhat recently, done offline. There are even utilities that turn GMail into a virtual hard drive. Instant messaging no longer requires the downloading and installation of software thanks to AIM Express and Google Talk. Securities trading can be done entirely without locally installed software through all major online brokers (this has been the case for some time). The only things the internet lacks which come to mind are the more computation intensive applications like media editing.

Highlighted by eyalnow

If there is a third-party intermediary, how can one be certain that the user’s information stays not only confidential and protected from prying eyes, but also that it remains privileged?

Highlighted by eyalnow

Additionally, when software gets centralized, it gets dumbed-down.

Highlighted by eyalnow

PS - I’m against this paradigm - if you depend on a central site then you have a single point of failure. If you have stuff distributed, you are safer.

— Posted by Ted KOppel

Highlighted by eyalnow

A central computing station can be located very near the source of power allowing greatly reduced electical transmission losses and/or free power for operation, as well as greatly extending the effective lifespan of equipment due to the reduced power costs.

Highlighted by eyalnow

Computing in the cloud means a TOTAL lack of privacy and a major blow to security. Not only will your data be stored in the cloud but even the process of synthesis of your raw data will be done elsewhere.

Highlighted by eyalnow

I think cloud computing advocates routine forget about the limitations imposed by the network - specifically with regard to latency. Applications that work very well in low latency environments will crash and burn in mid to high latency networks. This limits the appropiate applications to ones where the majority of the interactivity happens locally (like a webmail program) or where there is limited interactivity in the first place (streaming video). It doesn’t work for things where response times matter - games, command shells, and other interactive applications. The necessary delay imposed by the speed of light might seem small - whats a tenth of second between friends after all - but they can add up in ways that might not be obvious until the application is almost useless.

Highlighted by eyalnow

You can overcome these problems by pushing more and more of the program down onto the client *but* the more you have the client perform the further you move away from the thin client ideal. Eventually you end up with a situation where a ‘cloud computing’ device has as much processing power as a traditional system in order to handle all of the demands placed on it by the users.

Highlighted by eyalnow

When you couple this with increased network demands, the introduction of more failure points (the network goes down and your computer is now useless), the need to trust corporations with your private and confidential data (what happens when the next dot bomb hits and all of your data is on a server that gets reposessed?), and the lack of choice it entails I believe cloud computing will never replace tradtional computing. It may end up working within existing paradigms and compute infrastructure but it will never replace it.

Highlighted by eyalnow

all technical reasons aside they simply want the ownership.

Highlighted by eyalnow

Putting all data and increased functionality on the web is limited by: (1) the desire and need to work when the network is either unavailable or is functioning poorly; (2) the desire and need to maintain possession and control of one’s data and applications; (3) the loss of power and bargaining leverage to a third party that controls your data and your applications; (4) concerns about the security of your information on someone else’s servers; (5) the costs of renting or subscribing to services, which in many case will greatly exceed the costs of shrink wrapped software and the costs of owning and operating a computer; (6) Scaling servers and networks to run applications in addition to the current tasks of sharing files is going to be expensive, and that costs will be reflected in subscriptions for web services; and (7) for the foreseeable future, local applications on a PC are and will be much more capable and flexible than cloud computing for many applications

Highlighted by eyalnow

Cloud Computing shall compliment, not replace, local computing on the PC. The winner, whether Apple or Microsoft or someone else, will be he who best understands the constantly evolving principles of the relationship among cloud computing and local computing

Highlighted by eyalnow

Nothing discussed here is the real “cloud” - this just centralized computing circa 1970. The real cloud will occur when each of our personal computers is a low power networked machine/data storage, and the processing happens across massively multiple clouds of these low powered machines … the SETI distributed processing model as basic computing paradigm.

— Posted by Robert

Highlighted by eyalnow

For the discussion about thin client computing I strongly believe that centralized services and the personal computer will merge together in the future. That means that you can use centralized applications with local data and vice versa, and you do not need to know where the Application or data resides, you can simply use it from different systems and places.

Highlighted by eyalnow