Cloud Federation: It’s Happening; Business Models are Emerging
June 13, 2012Although far from a fait accompli, cloud federation is in the works. Like ants in the Internet ant hill, cloud providers are beginning to join forces to improve their individual prospects for long-term growth and sustainability.
Three companies — SpotCloud, OnApp, and Tier 3 — have placed bets on the idea that small-to-medium-sized cloud providers can benefit by federating resources. Each has a different business model that potentially blazes a trail for cloud federation. As its name implies, SpotCloud is a spot market that brings together buyers and sellers of commodity IaaS resources. OnApp federates cloud provider resources to deliver global content delivery network (CDN) services. And Tier 3 federates cloud provider resources to expand geographic reach and scalability for cloud computing services…
All three federation-enabling pioneers share two characteristics. Each has a proprietary cloud services software platform — and each has assumed the role of broker, bringing together suppliers and consumers of cloud resources and providing requisite usage tracking and billing.
SpotCloud was first to market a cloud federation-based service in 2010, when it introduced a brokerage service that enables IaaS providers to sell excess computing capacity, and buyers to find regional cloud providers for batch jobs. SpotCloud helps match supply with demand, applying flexible supply/demand-based pricing options. The service steps toward an auction-style marketplace for computing capacity similar to that for electricity.
According to Reuven Cohen, Founder and CTO of Enomaly, SpotCloud’s original parent company, SpotCloud has been slow to catch on. "The emergence of a federated cloud ecosystem is inevitable. The question is when. We received attention for being at the forefront of a market trend, but we are not where I thought we’d be by now. The need for a public market to provide equilibrium between buyers and sellers proved less than expected. But we did find an unexpected private market need; for large companies, governments, and universities to share capacity for private, not public consumption. That was our epiphany."
The University of California needed to share computing capacity among 38 campuses to more efficiently use resources amidst budget constraints. But each campus was a silo, and no system enabled them to share capacity. SpotCloud allowed capacity sharing using its marketplace model as a chargeback system. Governments and large enterprises also use SpotCloud to satisfy resource sharing needs. According to Cohen, "SpotCloud customers now have the ability to tap into the capacity that they have internally, and they also have a centralized way to tap into capacity with approved partners."
Cohen predicts that bona fide cloud federation will "pop" fully formed from what he calls "community exchanges". The community exchanges will grow and consolidate into public exchanges and marketplaces. He believes this metamorphosis will be sudden — like the adoption of social media, which as he says, "wasn’t until then all of a sudden it was". He also predicts that, like other resource exchanges, big financial institutions will play a major role.
OnApp launched its Federated CDN service to compete against Amazon’s CloudFront CDN offering in August of 2011. According to OnApp’s CMO Kosten Metreweli: "Given that we have over 400 service providers globally all running the same cloud platform, we realized that we could leverage spare capacity – some 20 percent of the infrastructure sitting there as an idle resource – and help cloud providers monetize that resource. We looked at how we could federate that resource to run services on, and the first of those was CDN. It was a natural fit, given the requirement to be in as many geographies as possible. So we built out a federation model to create a CDN application that could take advantage of spare capacity and create a marketplace that would allow [cloud] service providers to contribute their infrastructure and set their own price."
OnApp enables and operates the service, and is an intermediary and broker which tracks usage and manages financial settlements between resource buyers and sellers. Cloud providers deposit money into an OnApp CDN account, which is used to purchase CDN bandwidth from other Federation members. Revenue from sales of bandwidth is also stored in the account. OnApp issues monthly payments for any positive balance above the initial deposit.
Metreweli predicts that cloud federation will be analogous to eBay’s marketplace in that the winner will take all. "Once there is a marketplace that is successful, why go to another marketplace?" OnApp is betting that its CDN and follow-on services will bring ubiquity, scalability, reach, and more advanced service offerings to service providers, and by doing so, will reach critical mass to enable small-to-medium-sized cloud providers to compete successfully with Amazon. Although OnApp believes it can be the eBay of the cloud service provider world, Metreweli concedes that there is much work and education needed for that outcome. Luck will also play a role.
As an aside, OnApp faces a competing type of federation on the CDN front — that of network service providers who are itching to get a piece of the CDN action. In the past, Akamai and other CDN providers have pocketed the revenue. But increasingly, network service providers are realizing that they own beachfront property, and that a federated model is in their financial best interest. Cisco is fostering this effort through its Jet-Stream technology initiative, and Akamai is offering a proto-federation approach through its Aura Network Solutions offering.
Tier 3 began offering its Tier 3 Federated Cloud service to hosting and cloud service providers in February of 2012. A service provider hands over cloud service deployment and management to Tier 3. Tier 3 installs its Enterprise Cloud Platform software in service provider-owned, Tier 3-managed hardware, and configures the resulting "node" to work "in federation with" Tier 3-owned and other service provider-owned nodes. CFN Services, a company that provides automated financial trading services, is on board. According to Wendy White, Tier 3’s vice president of marketing, "In our model, CFN . . . provides [colocation] in an Equinix data center, and hardware — and we run our cloud software on top of that. They sell that node (and our other nodes) under the CFN brand and we sell that node under the Tier 3 brand as well." CFN is compensated for the use of its nodes using a revenue-sharing reimbursement model. One potential issue with the Tier 3 business model is that Tier 3 remains a cloud service provider itself, introducing a potential conflict of interest with its service provider customers.
Tier 3’s service provider customers benefit by expanding their service area. CFN Services touts this enhanced reach for its financial service-centric private cloud service. "Through a strategic partnership with Tier 3 . . . CFN is extending its existing Alpha PlatformTM, the company’s private cloud for ultra-low-latency market data delivery and trade execution, to enable financial services companies to self-provision an enterprise-grade, virtual private cloud for business-critical applications and services. The Alpha Platform On-Demand is the first public cloud tailored to global financial markets that is available beyond a single trading venue. CFN’s on-demand public cloud services will initially be available in North America across key liquidity venues in the equities, options, futures, derivatives and FX markets. In Q2, the Alpha Platform On-Demand will expand initially into European trading venues with anchors in London and Frankfurt and extend further into the APAC region later in 2012."
OnApp intends to add cloud service federation in future as Tier 3 is doing today, but it has chosen to lead with a CDN service as a more promising beachhead from which to federate cloud providers quickly. Unlike Tier 3, OnApp plans to remain a service enabler, operator, and broker — and not introduce a potential conflict of interest with its service provider customers by selling directly to enterprises.
What’s Next?
How and when cloud federation will reach critical mass is anybody’s guess — but the key to success will lie in the business model, not in standards and technology. In the end, standards such as OpenStack may play a role, but they will not drive cloud federation adoption because, as Reuven Cohen explains, "Money is the ultimate standard. You must either make or save money. If you don’t do either of those, you will not succeed."


