Enterprises have a complicated relationship with the cloud. Infrastructure-as-a-service (IaaS) offerings from public-cloud providers offer appealing alternatives to acquiring and provisioning on-premises hardware. And line-of-business organizations love being able to subscribe to software-as-a-service (SaaS) offerings that bypass IT altogether. But application development and deployment teams—the people the company charges with leading the digital transformation—have to work harder to gain the benefits cloud computing promises. And clouds add new facets to IT environments already struggling under the weight of too much of a good thing. But now, hybrid clouds—private, on-premises clouds linked to public clouds with data and applications shared among them—promise to take the enterprise’s love affair with cloud computing to a new level.
Descriptions of the cloud’s role in enterprise computing vary widely with who’s doing the describing. Public-cloud providers see almost all enterprise workloads moving to, yes, public clouds. To enable that transition, they’ve shored up their offerings with heightened security features. They offer service-level agreements covering availability and performance. And many now deliver a more complete application platform through platform-as-a-service (PaaS) offerings that bundle databases, development tools and other middleware on top of cloud-resident virtual machines.
But most enterprises are more circumspect. Shifting commodity applications such as web and email hosting from on-premises systems to cloud providers is usually straightforward. And shifting the accompanying maintenance and support effort may or may not save money, but it can free up people and funds for more important business initiatives. Public clouds also provide an attractive development and deployment platform for quick-start application programs, especially when their users are already in the cloud.
But special security and governance requirements, data-sovereignty dictates, exorbitant data volumes and the need for tight integration with other enterprise apps and workflows make many core business applications more suitable for on-premises deployment and operation. As a result, enterprises increasingly apply cloud technology to create private clouds—virtual compute, networking and storage resources running in their own data centers. Private clouds are an extension of their existing computing environments and leverage what businesses have already invested in virtual infrastructure and applications. Most enterprises see a mixed environment of public-cloud IaaS and PaaS platforms and private on-premises clouds for the foreseeable future.
To application teams, this situation leaves clouds—public and private—looking like additional deployment alternatives that must be developed for, planned for, configured for, tested for, automated, deployed on, supported and maintained. Development processes and DevOps deployment pipelines take more turns and lead down more paths. We can forgive them if their ardor for the cloud burns less brightly than in other parts of the organization.
But what if, to application teams, developing for and deploying to public clouds looked exactly like developing for and deploying to their on-premises private cloud? What if the same development and test tools and DevOps processes applied to both environments, so the app team could develop the application regardless of where it would run? What if selecting the deployment platform and deploying the app was a matter of simply making a selection in a tool at deployment time?
That’s the promise of hybrid clouds. And enterprise application teams are beginning to feel the glow.
Intel’s own IT organization is an example. It began building a private cloud in 2010. Initially providing internal IaaS services, it soon moved up the stack to provide PaaS and database-as-a-service (DBaaS). As team members gained experience, however, they discovered they had to shift perspective to think about services from the application down.
Using open-source software, Intel IT created an application platform that lets them develop, deploy and manage web and mobile applications. When creating applications, developers declaratively define the infrastructure needs of the app, then upload it through a web portal or interactive development environment to the platform. Intel IT’s Chris Sellers explains, “The underlying cloud platform abstracts all the infrastructure complexities for the developers including on-demand provisioning of resources, inter-component networking, patching and maintenance. This approach frees up developers and allows them to focus on what they are good at.”
This approach fits nicely with DevOps and continuous integration/continuous delivery programs, and it frees developers from managing—or even being aware of—the infrastructure environment. The actual deployment platform for each application can be selected on the basis of business needs, and applications can shift from one cloud to another if business needs or economics change.
After three years of operation, Intel IT has more than 3,500 application instances running in the new environment, and it’s in the process of rationalizing more than 2,000 enterprise applications to determine which ones should move to the new platform, which should remain in place and which can be retired or replaced. It can constantly reevaluate hosting decisions and shift applications as required to achieve the right balance of cost, features and functions. Their goal is 100 percent of applications running in the place that delivers the most business value at the least cost.
Although Intel IT’s application platform was an internal effort that relied on open-source software, commercial cloud service and software providers have also entered the arena—each from its own perspective and with the objective of making hybrid clouds easier for enterprises to create, use and support.
Microsoft is a public-cloud provider offering Microsoft Azure, so its solution extends Azure onto the customer’s premises. To create a uniform application environment spanning the customer’s private cloud and Azure public services, the company offers Azure Stack, Azure’s cloud application platform, but for on-premises deployment.
Microsoft works with hardware vendors—currently HPE, Dell, Cisco and Lenovo—to deliver Azure Stack pre-installed and integrated on validated systems. Alternatively, Azure Stack will be available as a managed service from select providers.
VMware is a software company, so its solution looks more like software. But it also works with hardware and cloud-service providers to create pre-integrated, verified configurations and managed services that ease acquisition and deployment. VMware Cloud Foundation builds on the company’s widely installed vSphere hypervisor to create a hybrid cloud platform that provides software-defined services for compute, storage, networking, security and cloud management. It can run enterprise applications—traditional or containerized—in private or public environments.
VMware offers VMware Cloud Foundation through pre-integrated systems from multiple hardware vendors, as software that customers can install on validated systems and as services in conjunction with VMware cloud providers including Rackspace and IBM Cloud. In addition, stepping into the cloud-service game themselves, they directly offer VMware Cloud services running on Amazon Web Services.
Red Hat builds on its open-source business model by offering Red Hat OpenShift Container Platform—a productized version of Docker-formatted containers and Kubernetes container-management software that gives enterprises a container-based application platform spanning development, deployment and operations.
Like other solutions, Red Hat offers OpenShift in multiple deployment models. The OpenShift Container Platform provides an on-premises software solution customers run in their own data centers. Other options include a managed cluster dedicated to the customer’s organization but running on Amazon Web Services or Google Cloud Platform and OpenShift Online, a cloud solution that Red Hat offers.
A common theme runs through these hybrid cloud solutions: ease of implementation through structured deployments on verified hardware and software platforms. That’s why the Intel Data Center Group has published more than 20 reference architectures to create alignment among various implementations.
Enterprises are learning the same truths about cloud computing they’ve learned about other transformative technologies: one size doesn’t fit all. It’s a hybrid business world out there, and hybrid clouds frequently offer the best fit. Creating one no longer requires a long journey through the open-source ecosphere. And the implementation alternatives offered by emerging providers promise to warm the hearts of CIOs, developers and operations teams alike.
Leading article image courtesy of Richard Patterson under a Creative Commons license
About the Author
William Giard is a Principal Engineer in Intel’s Data Center Group. William has over 20 years of experience in designing enterprise architectures and developing software solutions to support mission-critical systems across supply-chain, product-development and enterprise-infrastructure segments. Before joining DCG he led Intel IT’s software-development efforts to modernize the application and computing environment, delivering secure and usable solutions across multiple client computing platforms using cloud technologies to enable new business models.