How many Cloud Computing platforms would you say there are today?
Some abhor the notion of there being multiple clouds – by this thinking there is only one “cloud” in an almost Zen manner, meaning “the grid” and the ability to reach in, somewhere somehow, and use someone else’s compute capacity, web apps, services, storage, etc. Some others, however, as Amazon and others roll out their branded ability to do that reach, are beginning to call these “clouds” — I prefer to think about them as distinct platforms enabling cloud computing, but that’s starting to become a hazy definition.
Next week the world will hear more about Microsoft’s Mesh strategy.
I feel like an observer out on a prairie on a hot summer afternoon, watching the sky as cumulo-nimbus shapes emerge and burgeon across the horizon. The multiplicity is going to inevitably lead to feature differentiation, competitive marketing, a full hype cycle with naysayers and boosters (see Fortune magazine), down-market competition, shoddy wannabe clouds, boom and bust, market shake-out, etc. etc. – good times!
How many such platforms (how many clouds) will there be in future? How many should there be? And if multiplication really occurs, is this any different from “utility computing” and aren’t we heading back to the days of the mainframe-model of time-sharing?
Friend and former spook-agency colleague Bob Gourley wrote recently that he’s eager to check out Google’s App Engine, a platform aiming to enable cloud-based computing. He was unable to do so, however, because they’ve put a cap on the number of GAE participants, at least for now. When he is able to do so, though, he should think back to the days of the time-sharing model, for as Google warned when launching GAE, “Although Google App Engine will always be free to get started, we plan on allowing developers to purchase additional resources in the future, while paying only for what they use.”
I am interested in GAE and the variety of emerging cloud computing platforms (Microsoft, Amazon’s EC2, IBM’s cloud initiatives here and overseas, etc.). The key right now anyway is in the actual utility, meaning what can a user do with the cloud. (For example, I use Microsoft’s cloud every time I get an email in my Live account.)
Intrigued about GAE, I took a quick look around for GAE documentation, and noticed that one thing they have left open is the “GAE Issue Tracker,” a list of user-reported problems, bugs, and feature requests. An official Google blog characterizes the stream this way: “We’ve also seen a lot of feature and bug reports on our Issue Tracker,” and a review highlights where many of those bugs and limitations are occuring:
- If you want to use it, you better know Python. It’s the only developer language supported, which is quite a striking limitation in the broad world of web development. See here for a predictably heated discussion at Slashdot demonstrating the heterogeneity of that world. Perhaps at some point Google will add support for Ruby, Java, Perl, PHP, Fortran, etc, not to mention ASP and C#. Until they do I don’t know a lot of enteprise (even baby enterprise) users that will devote developer time to it.
- No support for offline processing – a function of the cloudness itself. GAE only works for web apps that engage realtime processing in response to user requests – no support yet for scheduled tasks or periodic (off-hour) larger-scale data migration.
- No support for large files – and by large I mean anything larger than 1MB. That limit holds for any particular request, both upload and download, which again is quite a crimp on usability. Both because of the limit and the previous bullet’s note on data migration, Google has already been forced to acknowledge to GAE beta users, “We know that many of you have large apps that already run on other platforms, and getting all of that data into the Datastore API is a challenge.”
Still, this model and Amazon’s model are of course more signals of the gathering clouds… hey, that could be a book title! Oops, too late.
PS: I like to tweak Google sometimes, they’re big enough to deserve it, but I hope no one takes this column as Google-bashing. If TechCrunch is correct, they accomplish that better on their own :-)
Filed under: innovation, R&D, Society, Technology Tagged: | Amazon, asp, Bob Gourley, business, C#, cloud computing, computers, computing, data, developers, ec2, elastic compute cloud, fortran, fortune, fortune magazine, future, futurology, GAE, Google, Google Apps Engine, IBM, IT, java, live mesh, marketing, markets, mesh, Microsoft, perl, php, programmers, programming, python, rails, ROR, ruby, saas, Salesforce, slashdot, SOA, software, tech, techcrunch, Technology, web services, windows live