When online marketing firm Hubspot started in 2006, the company’s IT needs were not very taxing, but they expanded quickly as the company realized success.
Hubspot – which provides software for businesses to coordinate their online marketing efforts – started with a private cloud for its front-end web hosting. Soon though, CIO Jim O’Neill realized he would need additional computing power for the company’s web analytics programs and “big data” storage needs.
The public cloud was a natural fit. “We really wanted the flexibility to scale up into a larger capacity when we needed it,” he said.
But the cloud environments had to work together, so O’Neill began the task of creating a hybrid cloud, which draws on the resources of both the public and the private clouds. For Hubspot, the hybrid cloud meant having the peace of mind of keeping important web hosting functions in its private cloud and the flexibility to use public cloud resources when needed.
The industry consensus, according to experts, is that most enterprises will eventually end up with hybrid cloud environments. But that’s easier said than done. “It takes careful planning,” O’Neill says.
A study conducted at Gartner’s 2011 Data centre Conference in December showed that 78 per cent of 2,500 attendees surveyed said they plan to build private cloud services by 2014. “We believe that the vast majority of private clouds will migrate or evolve into a hybrid model,” says Gartner researcher Thomas Bittman. “We’re telling clients to design and build their private clouds with hybrid in mind.”
The problem is there are no standards for interoperability between public and private clouds. So when enterprises go to integrate their clouds they face many challenges.
Hubspot’s O’Neill, for example, says the company had to resolve IP domain naming issues. When the company went to add public cloud support, O’Neill had to coordinate which IP addresses would be hosted by the company and which would be under the public vendor’s control. It took about a month of working with Amazon and Rackspace, the firm’s public cloud infrastructure as a service (IaaS) vendors, he says, but eventually the hybrid cloud was created.
While there are no standards for integrating clouds, Bittman says there are four “centres of gravity” for typical hybrid cloud environments, each of which have pros and cons in terms of facilitating the effort.
One of the centres is based on the dominant private cloud vendor, VMware, whose products enable enterprises to build an array of private clouds.
By its own accounting, VMware is a mainstay for more than 80 per cent of private clouds today, and while VMware does not offer public cloud services itself, it has a network of 94 partners that customers can work with to create hybrid environments.
That’s the upside. On the downside, it can be hard to integrate public cloud services from suppliers that are not VMware-centric, and Bittman says some enterprises fear building their whole integrated cloud environment around tools from one supplier.
A second centre of gravity, Bittman says, is emerging around the OpenStack model. Started by Rackspace and NASA in 2010, the OpenStack community has grown to include 152-member companies and the project’s goal is to create open source standards for cloud computing. Bittman says the project has the promise for creating a more friendly interoperability model for a wide variety of cloud vendors. On the other hand, Bittman says, OpenStack is still relatively immature and unproven. Only time will tell if it can serve as a key standard for building hybrid environments.
The third major centre is the dominant public cloud IaaS provider, Amazon, which offers a series of cloud options through Amazon Web Services (AWS). These range from its Elastic Compute Cloud (EC2) to its Simple Storage Service (S3) products. The company is working to support compatibility between its public IaaS offerings and the private clouds enterprises already have, Bittman says. Amazon publishes, for example, its APIs, which simplify the job of integration, he notes.
And finally the fourth centre of gravity encompasses most all of the other providers, Bittman says. Large vendors, such as IBM, HP, Intel, Microsoft and others, each have their own set of cloud offerings, ranging from Microsoft’s Windows Azure platform as a service (PaaS) model, to IBM’s SmartCloud, which includes both IaaS and PaaS products. Some of these players, including a growing number of telecommunications carriers such as Verizon, CenturyLink and AT&T, offer a variety of both public and private cloud services, and are making it easier to create hybrid clouds.
AT&T, for example, this month released its Synaptic Compute as a Service, which makes the company’s IaaS public cloud compatible with VMware’s vCloud Datacentre offering. Recognizing the integration challenge ahead, many vendors are acquiring assets to help ease the transition. Last year, for example, after purchasing IaaS provider Terremark, Verizon bought cloud integration specialty firm CloudSwitch, and IBM bought Cast Iron, another integration company.
So which camp should you bet on if you see a hybrid cloud in your future?
“It’s really about answering the question: ‘what problem are you trying to solve?'” says Toby Owen, senior manager of hybrid cloud solutions for Rackspace. If it’s to transfer and store data, for example, then questions about bandwidth and data transferring could be top of mind. On the other hand, if an enterprise is using the cloud to support bursting, or a sudden spike in web traffic, perhaps price is a more important consideration.
Whatever sways the decision today, it is important to consider that needs may change in the future. Which is why, Bittman says, it is so important to consider interoperability issues going in.