Editor's Note: Part two of our three-part series on the business and technology essentials for building out a cloud services infrastructure looks at the technology options available to cloud providers to enable service deployment and management for their customers. After deciding which services they plan to offer, providers need to identify potential suppliers that will help them build the required infrastructure. This technical overview helps you analyze the requirements for your cloud delivery network and data centers, then looks at cloud platform options to enable applications and services. As an added benefit, check out the list of questions you should ask your vendors before making a product or service choice. And, finally, as an added benefit, have a look at the list of vendors competing in the cloud market.
To enter the cloud services business successfully, cloud providers must match market expectations to cloud infrastructure through their service definitions. Market expectations for cloud services are usually framed in terms of the type of cloud offering -- Infrastructure as a Service (IaaS), Platform as a Service (PaaS) or Software as a Service (SaaS).
Since the universal goal of IT is to run applications, IaaS and PaaS cloud services still require organizations to contribute and support some components, including the applications themselves, operating systems, hardware platforms and middleware. Higher-level cloud services like Software as a Service displace most enterprise IT costs by including the applications, and thus command the highest prices and profit margins for cloud providers.
Lower-level services like IaaS are more easily adapted to applications that are already owned and operated by the enterprise, so they are often easier for cloud providers to sell. In the early cloud market, IaaS has dominated as the lowest-hanging cloud service fruit, but most research indicates that the majority of the cloud opportunity will be addressed with PaaS or SaaS.
Building a cloud services infrastructure
Once an operator has determined which type(s) of cloud service to offer, the next step is to frame the infrastructure required for deployment and management and identify its potential suppliers. A prospective provider has a variety of options, including:
- building its own cloud infrastructure;
- adapting current network and computing facilities for cloud services; and
- buying wholesale components of cloud service from others.
Generally, a cloud provider will need a specific value-add to make its service competitive and to build its own profit model, so even if some service components are bought wholesale, the prospective provider will probably have to augment them with custom cloud elements.
Vendors at a glance
Delivery network and data center requirements.
A cloud infrastructure starts with a data center and a delivery network. The data center provides servers and storage, and the servers in the data center run a cloud platform, which is a set of software tools. These tools manage the compute and storage resources of the cloud, allocate resources to users, manage the resources and user relationships, secure the infrastructure, and perform billing and accounting for resource usage.
The cloud delivery network starts with a data center switching network for server/storage connectivity, then adds the application and the ability to steer the user to the correct cloud resource to access applications (since there are no hard server assignments). Finally, it links the result to a wide area network (WAN) service, which can be the Internet or a private network like a VPN. Note that some cloud operators are willing, for a fee, to connect their clouds to a legacy technology like frame relay rather than a VPN or a VLAN.
Choosing a cloud platform.
Cloud platforms vary considerably, not only in the specific features for resource allocation and management, but also in the type of application and service mix they are designed to support. In theory, SaaS services can be offered from any set of servers, operating system (OS), and middleware that supports the application software. For PaaS and IaaS, the cloud platform must explicitly support the service offered. Currently, IaaS services are most often created using a server farm equipped with virtualization. PaaS and SaaS services can be built on top of this framework, as well. PaaS services can also be built using a vendor-specific cloud platform like Microsoft Windows Azure, for example, or based on open cloud software like Hadoop.
Remember that the chosen platform may limit the applications and services that can be offered, which makes it critical that business goals be carefully matched to the platform. When in doubt, it's best to take the most general approach possible.
Location and partnerships.
Another factor to consider in selecting cloud components is the proposed service geography and potential partnerships or other wholesale relationships that may be needed. Other relationships are needed to either cover areas where providers can’t economically build their own facilities or provide early expansion capacity if the cloud service opportunity out-paces the infrastructure buildout.
If you expect to use wholesale or partner interconnections (often called cloud federation), it is important that your own cloud infrastructure matches or is at least compatible with that of the wholesale and partner resources. Some cloud platform providers offer their own public cloud services, and using these may be a good way to expand your geographic scope and add capacity to a cloud service.
Single-source cloud-in-a-box options.
When selecting cloud platform and network components, buyers can try to find a single source, a "cloud in a box," or select specific parts of their configuration based on service needs. In that case, some integration of the elements will be required, of course. You can ask one of your major component suppliers to act as an integrator, do the integration in-house, or hire a third-party firm that specializes in cloud building. All three choices will be available for nearly every cloud-building project, and it's important to look at the total cost of each, including not only vendor charges but internal costs, too. Companies lacking in cloud expertise will likely find self-integration too risky, for example.
Tracing business goals to available features.
When cloud service objectives have been aligned with an X as a service set, and when specific platform and network elements have then been identified, it's time to look at the available products and services and make an optimized selection. There is no simple blueprint for selecting the features of cloud infrastructure because the requirements will depend on the type of service being targeted and the nature of any pre-existing hardware or software used as a starting point for offering cloud services. The key to success is to trace every business goal through a service capability and then to a software/hardware component. This will establish the requirements for every component, and operators will have to ensure these requirements are met by the selected elements of cloud software and hardware that the operator will use to build a cloud service offering.
When emerging cloud operators have a service strategy and an inventory of the components needed to support it, they can begin to examine the potential product and service providers that could be the best fit.
Learn more about Tom Nolle, President of CIMI Corp.