BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
Businesses that are hesitant to shift to public clouds may find that a hybrid cloud provides the right combination of internal control and public cloud benefits. Cloud computing has matured to the point where we have a variety of public and private cloud options, so determining the right components for your hybrid cloud requirements is a challenge. Here are five key topics to keep in mind as you evaluate your options.
Interoperability. The purpose of a hybrid cloud is to have the option of moving workloads from your private cloud to a public cloud when needed -- in case of, for example, an outage in your private cloud, peak demand for computing resources or software development testing. The cloud platforms used in your private and public clouds must be interoperable to ensure low-friction movement of workloads between clouds.
Interoperability is achieved by running either the same cloud platforms or platforms with a shared application programming interface (API). OpenStack, for example, is an open source cloud framework available for private clouds. A number of public cloud vendors, including Rackspace, IBM and Hewlett-Packard, use OpenStack. On the other hand, Eucalyptus is API-level compatible with Amazon Web Services' proprietary cloud services platform.
If your requirements are not met through a common or interoperable cloud platform, consider a cloud infrastructure management service, such as Enstratus or Rightscale. These services offer features like consolidated management and billing, as well as support for open security standards. You will incur additional charges using these services, but the benefits of consolidated management and billing, along with the ability to work across a variety of clouds, can outweigh the costs.
Security. Security is frequently cited as a reason to not use public clouds for some applications. If you extend your IT services to support a hybrid cloud, consider extending your virtual private network (VPN) to the cloud. A VPN will enable your systems to communicate over encrypted channels across clouds.
How do you consolidate identity and access management (IAM) controls across private and public clouds? A single set of access controls and policies will help mitigate the risk of discrepancies between the clouds. If you are working with different but interoperable cloud frameworks, assess how well your policies extend across clouds. Interoperability is a spectrum and a common API for some functions, such as managing instances, but it does not guarantee a common API for others, such as security functions.
Cloud management. A unified management system can streamline routine tasks working within a hybrid cloud. Ideally, a management system will have a consolidated view of private and public compute and storage resources as well as consolidated billing. Management alerts should take into account reource use in both public and private clouds. If a department sets a budget limit for cloud resources, managers should be able to receive alerts when the combined expenditures for both clouds approach that limit.
Data management. High-performance computing (HPC) teaches us that data should be kept close to the computing device. This reduces the time required to transfer data from storage to the server and reduces the risk of conflicts at a centralized data store. A similar principle is at work with hybrid cloud computing: Data should be located in the cloud that will perform the majority of the processing on that data.
It is relatively easy to shift processing among clouds. Copying a configuration script among clouds is trivial and copying a machine image is within reason. But copying large volumes of data among clouds can be time-consuming and, depending on your public cloud provider's pricing model, expensive. Develop guidelines for data management to help developers and application administrators avoid unexpected charges or long delays when migrating tasks between clouds.
Service-level agreements. Both the public and private cloud providers should have established service-level agreements (SLAs). Availability of computing and storage services, durability of data and prices should be defined in SLAs.
Different use cases may be affected by an outage in one of the clouds. Application owners responsible for systems that can only run in the private cloud may expect higher levels of availability and greater compensation for outages than are offered to users who can shift their workloads to the public cloud.
About the author:
Dan Sullivan, M.Sc., is an author, systems architect and consultant with more than 20 years of IT experience. He has had engagements in advanced analytics, systems architecture, database design, enterprise security and business intelligence. He has worked in a broad range of industries, including financial services, manufacturing, pharmaceuticals, software development, government, retail and education. Dan has written extensively about topics that range from data warehousing, cloud computing and advanced analytics to security management, collaboration and text mining.