Prepare Ye the Way of the Intercloud
January 9, 2012
The term Intercloud describes the future interconnectedness–or federation–of clouds, similar to the network of networks that is today’s Internet. Like the electrical grid, in which a utility is shared based on supply and demand, the Intercloud will be a mesh of cloud computing resources owned by many, and interconnected and shared via open standards.
Economic forces are propelling cloud service providers toward peering arrangements that presage cloud federation. Federation allows cloud providers to buy external resources when demand exceeds supply, and to rent resources when fellow providers need help meeting demand. It also allows providers to offer service outside their footprint without deploying computing resources in every nook and cranny of the world. Sharing resources among providers is especially compelling for smaller service providers because it enables them compete on a more even footing with Goliaths like Amazon’s EC2…
Realizing the Intercloud vision requires overcoming formidable technical and organizational challenges. While some bailing-wire-grade software solutions exist to help cloud operations work together, the underlying cloud federation blueprint remains an academic exercise. Scholars from Australia to Italy are racing to create theoretical frameworks that lay out the functions needed to enable the Intercloud, and they have crafted exquisite models for how to cobble these functions together into a working system.
On the operations side, both open source and proprietary software solutions have already begun to hit the streets. Work under the open source OpenStack initiative launched by Rackspace Hosting and NASA in July of 2010, has resulted in software and standards for large-scale deployments of automatically provisioned virtual compute instances applicable in a cloud federation environment. The OpenStack project aims to enable any organization to create and offer cloud computing services running on standard hardware. CA offers a proprietary, turnkey system called 3Tera AppLogic that converts arrays of servers into virtualized resource pools.
On the management side many as-yet unmet technical challenges loom. Service load must be dynamically coordinated among cloud service providers to determine where best to host for optimal performance. A common authentication scheme must be implemented. Data and application security must be assured. Secure connectivity between clouds must be assured. And reconciliation and billing must be implemented to properly bill and compensate participating cloud service providers.
Some of these federation management challenges can likely be worked out quickly on a bilateral basis between individual cloud service providers. This will take effort on the part of service providers, vendors, and partners–and those efforts have begun.
Fully realizing the Intercloud vision, however, will be a much more complex, long-term, and collaborative undertaking. To give you a sense of what is involved, a recent scholarly paper by Rajkumar Buyya, Rajiv Ranjan and Rodrigo Calheiros at the University of Melbourne and the University of New South Wales in Australia describes an Intercloud system architecture encompassing an array of functions and players that do not exist today. These include: a cloud exchange composed of directory, auctioneer, and financial settlement functions; cloud coordinators to do scheduling and resource allocation; and cloud brokers to negotiate with cloud coordinators for resource allocation that satisfies users’ QoS needs. Within each function are many sub-functions, which also do not yet exist. This paper is but one of many on the subject, each putting forward a different approach. Place your bets ladies and gentlement.
Some current solutions show promise as pieces of the overall cloud federation puzzle. For example, the German cloud service provider ScaleUp Technologies claims their ScaleUp Cloud Management Platform designed that provisions and manages technical aspects of cloud service delivery, now supports federated environments. And a set of solutions from ComputeNext promises to: "match resources across heterogeneous data center environments and seamlessly federate workloads so that service providers can take the best workload (based on capacity and margin) for their businesses and continue to offer the best services to their customers."
Common authentication solutions for cloud federation are also in the works. For example, SinglePoint Trust CloudTM by Symplified is an identity, access management service designed for cloud federation environments. Radiant Logic’s RadiantOne Cloud Federation Service, "enables a secure federated infrastructure by supporting claims generation for all uses in diverse and disparate data stores." And RSA is in the game with RSA Cloud Trust Authority, which provides cloud-based services for identity, information, and infrastructure cloud security to secure interaction among cloud service providers. Not to be left out, Microsoft is pushing Active Directory Federation Services as the foundation for identity in cloud computing. Last, but not least, Cisco is bandying about the Intercloud term, but so far it does not have specific offerings to support cloud federation.
In our view, it is time for standards bodies and industry consortia to hammer out the architecture and standards underlying cloud federation. Identifying and agreeing upon the requisite function set, and building and deploying the interconnecting pieces will be neither quick nor easy, but the it will be worth the trouble-and the results will spark innovation and economic opportunity for forward-thinking companies. Although just a glimmer on the horizon, the Intercloud will come into its own.