Read the Original Article at http://www.informationweek.com/news/showArticle.jhtml?articleID=240002733
But both Microsoft and Google have always had the means to compete in that market--they've built data centers in strategic locations around the world and can use them as a springboard into IaaS. And some customers want Microsoft and Google to become IaaS suppliers, given their existing reliance on those vendors' technologies.
In fact, both have been edging closer to doing so.
Google's primary infrastructure architect Urs Holzle on Thursday at the company's developers conference announced the Google Compute Engine, or infrastructure-as-a-service, a move that brings it into more direct competition with market leader Amazon Web Services.
Holzle said Google's success with App Engine, its platform-as-a-service for building applications, had generated demand among its users for an accompanying infrastructure service to run applications in virtual machines for long periods of time. Google will manage all scalability and performance issues for users of Compute Engine, letting businesses and independent software developers gain the benefit of Google's "decade of experience in building and running Google search engine data centers." It made the announcement during a Thursday morning keynote session of Google I/O in San Francisco.
Google already has one the largest worldwide networks of data centers with strong networking connections between them. That may become a distinguishing point between Google and Amazon, which also has data centers around the world but is just starting to offer connections between them. Based on its approach to platform-as-a-service, Google can be expected emphasize the resiliency and lack of downtime of its approach to infrastructure.
Amazon, however, offers infrastructure for both Windows and Linux workloads. In addition, Google Compute Engine will host virtual machines running only Linux workloads.
Holzle said Compute Engine will deliver "up to 50% more value" than other IaaS providers without specifically naming any competitors. In addition to Amazon, Rackspace, SoftLayer, and Hosting.com offer IaaS, along with more business-oriented suppliers such as Terremark, a unit of Verizon, and Savvis, a unit of CenturyLink.
Compute Engine is available immediately in "limited preview," meaning it doesn't yet qualify as a full-fledged product. Google will limit how many users may sign up for the service. No date for general availability has been named.
One of the strongest steps Microsoft has made toward IaaS came June 6 whenit said it would start offering Linux servers. If it's ever going to offer IaaS, it needed to include Linux or permanently concede a large part of the market to Amazon. It will import the Linux workloads using its VHD virtual machine file format, meaning that its Hyper-V hypervisor will gain a more equal footing with VMware's ESX Server, which from day one dealt with Linux.
Microsoft at the time said that Azure customers may run CentOS, Ubuntu, or Suse Linux servers--in other words nearly anyone except Red Hat. Red Hat is still too formidable a competitor. Despite being a denouncer of Linux for years, Microsoft has reconsidered. It doesn't wish to be left behind because of its militant line. If cloud users want to run Linux, Azure will do so.
Why Open Source?
[ There is a lot of moving around in the cloud. Read NASA Drops OpenStack For Amazon Cloud. ]
If cloud users find in Azure what they need to build their websites, they're more likely to use it for its infrastructure and build more applications using its platform-as-a-service. PaaS is a step above plain vanilla computing because it adds tools, widgets, and frameworks for rapid application building; the resulting apps run conveniently in the PaaS cloud.
Thus, Microsoft has maneuvered itself much closer to becoming an IaaS provider. It has invested heavily in data centers around the world--Azure will become available in 48 additional countries, including Russia, by the end of June, bringing its total to 89. Microsoft is thus in a strong position to make an IaaS offering as a logical extension of its platform. Offering IaaS would also allow Microsoft to more thoroughly utilize its data centers for more profitable operations.
But so far, Microsoft hasn't gone head to head with Amazon, except when offering entry-level small servers. Both Amazon and Microsoft appear to be trying to capture the cloud newcomer and early cloud user with low pricing for small servers. As the size of the server offering increases, Microsoft loses a small pricing edge and Amazon offers the stronger IaaS server resources for the money. In that sense, Microsoft appears to be competing for IaaS customers with the intent of leading them toward a value-added and higher-priced PaaS environment.
Google, on the other hand, keeps expanding how its App Engine works with the outside world. It may not be trying to bring its primary compute workloads into App Engine, an environment that runs Python and Java (or languages, such as JRuby, that run in the Java Virtual machine). Rather, it appears to be matching up App Engine with third-party services that can bring it on-premises workloads. That would allow App Engine to function as an extension of customer's infrastructure without needing to transform into general-purpose infrastructure on its own.
App Engine, like Azure, is also primarily a PaaS platform, with Google enhancing it with Google Gadgets, Google Apps, and other services. Its just-announced partnership with Cliqr Technologies is aimed at customers who are creating on-premises clouds and looking for them to interoperate with a public cloud.
Driving A Hybrid
Interest in hybrid cloud computing, where workloads may run either on premises or in the public cloud, is growing. Cliqr makes it easier for them to move those private cloud workloads somewhere else. In entering a partnership with Cliqr, Google is positioning App Engine as a destination for those seeking operations in the public cloud.
Google has restricted itself to offering a compute platform highly compatible with the way it likes to produce applications. In a sense, it's offered its spare capacity as a compute platform and provided tools it built for its own use to the rest of the world.
Both Microsoft and Google PaaS could migrate into something big, a direct head-to-head competition with Amazon to provide general purpose IaaS. Then again, the principals may be engaging in evolutionary baby steps, afraid to do a self-transformation through a major DNA transfusion.
Google's cloud environment started out able to host applications written in the language its developers most frequently use, Python, a high-speed language in which programs may be quickly modified. It later expanded App Engine to be able to run Java or the other languages geared to run in the Java Virtual Machine, such as JRuby. But Google did not produce a plain vanilla, "send us anything to run and we run it" platform.
In Microsoft's case, it looks like IaaS, with the ultimate goal of moving the customer to PaaS, including some Windows parts and products. The way it implemented Linux support suggests that it wants to keep developers engaged with its platform and is making it easier for them to move a Linux project into the Azure setting.
In Google's case, it looks like hybrid cloud operation--a customer on-premises system that works with Google's public cloud--to bring more business to Google App Engine. Google has not only built out an extensive chain of data centers but has learned to minimize user response times through experience with its search engine.
Rent The Spike
Cliqr is a Google Ventures and Foundation Capital-backed company that can give a hybrid cloud user the means to view workloads running in either setting and project their expense. Cloud users such as online gaming company Zynga are major proponents of the hybrid approach. Companies should only build enough data centers to run their applications in steady state, off-loading heavy traffic demand into the public cloud. Zynga's cloud architect Allan Leinwand sums up the approach as, "Buy the base, rent the spike."
Other companies may wish to offload public-facing functions as website operations and marketing promotions to the cloud because little sensitive corporate data is involved. They would maintain in-house the jobs that require strict data privacy and compliance.
Cliqr is a startup that aims to bring in house the operational smarts now found in outside third parties, such as RightScale, which can move its customers' workloads around from cloud to cloud, handling the compatibility issues within its automated processes. Cliqr would perform a similar function for App Engine, mediating between an on-premises workload and its export to the public cloud, where its console can help manage it.
Expertise, automation, and silo busting are all required, say early adopters of private clouds. Also in the new, all-digital Private Clouds: Vision Vs. Reality issue of InformationWeek: How to choose between OpenStack and CloudStack for your private cloud. (Free with registration.)