Google Tackles Microsoft with Office
By taking its most emphatic step so far towards a future in which web services supplant most of today’s software, Google last week sent its clearest signal yet to Silicon Valley and Wall Street about where it is heading.
The test version of Google App Engine allows third parties to develop and deploy web applications that run on top of the vast global infrastructure Google has spent billions on to support its search and advertising business.
The third-party apps will reside on Google’s estimated million or so servers and use APIs to access and make use of the software it uses to monitor, manage, secure and scale its cloud computing environment.
As Google has made so much noise about the power and scalability of its infrastructure, and its eagerness to attack Microsoft’s lucrative applications businesses through low-cost web alternatives, it is perhaps surprising that it took so long to open access to its platform.
Its resources and reputation, plus the current mania for Web 2.0, ensured that it took only two hours to fill 10,000 slots allocated for outsider developers to use Google’s beta platform free of charge.
If there was one scary aspect for Microsoft it was that, until now, no competitor had approached its ability to cultivate third parties, offer tools to simplify software development, and ensure they contribute to an system that makes Windows more attractive and hard to displace.
Microsoft is expected to unveil its own .NET cloud services for external developers in coming months, but has decided to delay announcing them until they are generally available, rather than in beta.
Cloud computing - shifting complex computation from individual servers or desktops to a distributed, shared, fault-tolerant pool of computers accessed over the network - has been a vision across much of the IT industry for decades. Sun famously built its business (or at least its advertising) on the premise that the network is the computer.
Until recently, the vision remained largely just that. Only over the past few years has it solidified, thanks to faster computers, always-on connections, smarter synchronisation and caching software, and the emergence of web standards.
It has been given fresh life by the vast piles of hardware that big internet players - most of all Google, but also Amazon, Yahoo, MSN and others - deployed to run their businesses.
Since their systems must handle peak loads, some of this infrastructure is idle much of the time. These firms have also had to develop capabilities in managing complex distributed applications (instant messaging or photo sharing, for instance) cheaply with minimal downtime - a task that requires automating routine tasks and component failovers, controlling energy use and so on.
It wasn’t much of a leap from there to the notion that these firms could sell (or rent) access to spare resources (computing cycles, bandwidth, and database and disk capacity) and their hard-earned management and automation skills.
Amazon was the first to do so via Amazon Web Services, an offshoot that sells a range of services housed at the firm’s data centres to thousands of paying customers, many to back-end Web 2.0 applications.
At one level, Google App Engine overlaps with Amazon’s cloud services, and this was the main reaction to its release, but the parallels should not be overstated.
Amazon offers mix-and-match computing capabilities, and leaves it to users to decide which to combine. This flexibility means almost any app - not just real-time web applications - can use the platform.
Google App Engines, on the other hand, is a narrower and tightly bundled set of services that provides end-to-end support for rapid development and deployment of web applications (and only web applications). It is possible a web app built on Google’s front end could use Amazon as back-end storage.
The two overlap in intent, particularly by allowing third-party users to piggyback low-level services and management frameworks. Both Google and Amazon aim to abstract them from what developers call yak-shaving - duplication of low-level, often obscure functions only distantly related to the app actually being built, but needed for it to work.
By providing pre-shaved yaks - standard low-level services called through an API - the Google App Engine in theory allows outside developers to avoid using yet another queuing or routing routine, and concentrate on high-level user-visible attributes of their project. That is its great promise.
Google’s first move into this space is certainly not perfect: notably its initial version, and SDK supports only one programming language, Python, a scripting language widely used in Google and elsewhere, but far from universal. Google will add other languages over time. An unspoken issue in all this is how easily an application built on the Google platform can be repurposed to use a cloud offered by a rival vendor.
Google seems to be trying to make it possible for users to not be entirely locked in (in contrast to the approach of Facebook, for instance, which may feel some heat from this move). True portability is not in any vendor’s true interests, and will likely remain elusive.
Still, as a shot that finally poses a real challenge to Microsoft (and, at a different level, Facebook) the Google App Engine is a major move, and at last offers a clearer indication of a highly profitable “beyond web search” element of the company’s bigger picture.
Loading ...
[...] [...]