While the IT industry, analysts, and media continue to debate, define, and document cloud computing, enterprise customers are already testing and deploying private clouds. By studying these successful early deployments, software vendors can better understand the key elements necessary to build and manage private clouds.
As a result of my own conversations with customers, industry analysts, and partners, seven discussion points consistently surface as necessary elements to evolve internal infrastructures to a private cloud. This short list provides software vendors with insight on how to counsel their clients about the capabilities needed to successfully manage a private cloud environment.
Enabling private clouds
Cloud computing is an IT delivery model that offers large-scale, shared infrastructure and computing resources as a service through self-service, pay-per-use access. It is the logical next step in the evolution of the data center from mainframe and now to clouds. Although cloud computing leverages recently developed technology, I believe that cloud computing is a business – not a technical – trend.
Private clouds are used by organizations to pool internal IT resources for shared use and enable managed access to external IT resources. One thing that sets private clouds apart from public cloud infrastructures is that they are owned and managed by an organization’s IT team, so the control and security of the infrastructure is handled internally. Private clouds also support IT’s obligation to oversee corporate requirements including governance, compliance, business continuity, cost reduction, and risk management.
By leveraging internal IT data center resources, the private cloud maximizes an organization’s existing infrastructure by pooling legacy systems and resources for self-service reuse across an organization. One key component of private clouds is management software, which acts as the control layer needed to pull resources into a unified cloud environment and deliver them as services.
Management software is the critical enabler for IT to configure its existing data center infrastructure, integrate its management tools, and support its applications and business processes in evolving to a shared internal infrastructure. Just like other mission-critical enterprise business systems and services, private clouds are usually custom-built by the IT organization, not bought from a vendor, to ensure they conform to the unique computing needs and policies of the organization.
It’s also worth noting that on occasion, a private cloud infrastructure may incorporate resources from external service providers or the public cloud, which is usually referred to as “cloud bursting.” This can help avoid the necessity of provisioning for peak aggregate resource demands.
Seven private cloud components
As software vendors work with enterprise clients on how to best evolve their internal infrastructures to a private cloud, here are seven key components to keep in mind:
- Heterogeneous systems support – The private cloud must support an organization’s heterogeneous infrastructure, as well as resources from external providers. This includes server, storage and networking hardware, operating systems, hypervisors, storage systems, and file systems.
- Integration with management tools – Enterprises use a variety of IT management tools for security, provisioning, systems management, directory, reporting, billing, data management, regulation and compliance. Cloud computing does not replace these tools, and companies should not have to buy all new tools to work with their cloud. Instead, properly designed private cloud management software easily integrates with existing tools and invokes them as needed during cloud operations.
- Configurable resource allocation policies – The cloud must be workload-aware as well as resource-aware. This means that your cloud management software should determine the most efficient placement of user requested workloads. This guarantees resource reservations to its customers based on well-defined policies. And, when demands peak, you can automatically arbitrate resources based on business priorities of various parts of the cloud workload to cost-effectively meet SLAs.
- Integration with workload managers, middleware and applications – Clouds exist to run applications. In addition to having a self-service portal for users to request virtual or physical machines, you need a flexible programmable interface to enable easy integration with the enterprise’s essential workload managers, middleware, and applications.
- Support IT and business processes – Clouds provide support for various IT and business processes and allows IT to automate many of its operations. The self-service aspect of cloud computing empowers users to request and obtain physical servers and VMs in minutes instead of days or weeks. Cloud management enables the definition and ongoing modifications of many IT management processes that previously would have been performed manually.
- Extensible to external resources – In addition to providing more flexible services with internal resources, the cloud should enable managed access to external resources that are hosted by service providers. This allows for more flexible capacity planning where additional resources can be used and paid for only when needed, while centrally controlling access and metering of these services.
- Enterprise, not workgroup, solution – An organization usually consists of multiple departments and locations, often distributed internationally. A flexible cloud scales to meet their diverse needs in real time. While cloud computing may be adopted initially within an individual line of business or location, it enables the integration of IT across the enterprise by reconfiguring rather than replacing the private cloud management software. Therefore, a private cloud can be an enterprise-wide IT services delivery system that provides transparent and consistent access to global resources.
The decision to evolve an organization’s infrastructure to a private cloud needs to be evaluated closely to ensure that the private cloud that is built – not bought – and supports specific business goals and IT needs while capitalizing on an organization’s heterogeneous IT infrastructure.
Martin Harris is Director of Product Management at Platform Computing.