Legos

As composable infrastructure emerges, the data center evolves

Until recently, "simple" and "effortless" were two words rarely used to describe an enterprise data center. With the increasing focus on agility in today's enterprise software-defined data center, however, that is changing.

Enterprises are looking to the public cloud for the agility they need to thrive in the digital economy, where applications based on mobile and cloud technologies generate more and more revenue. (Mobile apps alone are expected to generate $77 billion in revenue by the end of 2017, according to Gartner.) DevOps practitioners tend to associate the public cloud with greater automation, simplicity, and convenience to stand up resources for the next-gen apps they need to create and run, even though the costs of running data and applications in the public cloud can rise quickly.

At the same time, running traditional ERP and database applications on-premises—along with some newer apps where data ownership, security, and compliance are concerns—comes at the potential cost of over-provisioning for peak loads.

Here are several ways that a software-defined data center can address these issues.

Best Practices Guide: Migrating Applications to the Cloud

The rise of flexible infrastructure

"Customers tell us all the time that they're dealing with two worlds," Ric Lewis, HPE's senior VP and general manager of converged data center infrastructure business, said during a session at Discover Las Vegas in 2016. He noted that CIOs are frustrated by recommendations to maintain these separate cloud and on-site divisions. "They say, 'Why can't I just have infrastructure that's reliable and that flexes to the needs of the application?'"

The concept of the software-defined data center has been evolving to fill that need, said Lewis, most recently with the added refinement of composable infrastructure. A composable infrastructure is a framework that allows organizations to treat the traditional resources of a network, including compute and storage requirements, as services via APIs. 

Building out a software-defined data center

Enterprises need an intelligent, intuitive, and dynamic software-defined infrastructure (SDI) that can flex to the needs of any workload. That may be a traditional database running in a virtual or bare-metal environment, or a containerized back-end app that uses the open-source tool Docker.

As of 2016, only 21 percent of enterprises had implemented a software-defined infrastructure, according to a survey of more than 900 IT professionals conducted by 451 Research. The report notes that, for senior management, low product maturity is a top barrier to adopting SDI, which encompasses everything from virtualizing hardware resources to automating server, storage, networking provisioning, and elastic infrastructure scaling.

One takeaway is that, to date, most businesses can't deliver a complete public cloud–like experience from their data centers. Which means they can't fulfill the "infrastructure as code" enticement that makes the public cloud attractive to DevOps advocates. Beyond this setback, IT can't apply the infrastructure-as-code model to reduce costs for traditional applications by reducing overprovisioning.

That's changing, however, thanks to composable infrastructure. Here, compute, storage, and network fabric come together in one platform, reconfigurable through software-defined intelligence to become whatever is needed—whenever the enterprise needs it—in any kind of operating environment.

Composability defines the new SDI

New-world IT applications can draw upon fluid resource pools of self-discoverable, self-identifiable, and self-assembled components, provisioned with a single line of code, to meet their infrastructure requirements. Spare capacity can be instantly available across all data center applications, reducing operational costs for all IT systems, too. Resources will never be left adrift, as they can go from supporting a legacy application one minute to being recomposed for a new-world IT application the next.

"What it's really about is delivering dynamic infrastructure that can flex to the needs of a workload," Lewis said. It's the answer to the CIO's desire to use a single operational model and infrastructure stack for all workloads and operating environments, providing public cloud–like agility from in-house, software-defined data centers.

Lewis described a senior IT architect at a satellite TV provider who has been trying out composable infrastructure technology to draw on the dynamic, fluid nature of software. The concept takes the need for infrastructure provisioning "into the hardware layer, where now we don't have to worry about, 'Oh, do we have to get the right hardware for that project, for that application?' ... It's flexible. That's the big thing," said Lewis.

Best Practices Guide: Migrating Applications to the Cloud
Topics: IT Ops