Like a childhood nickname that just won't stick, composable infrastructure has yet to resonate with many data centers...
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
-- but that might be changing, thanks to DevOps.
IT pros have heard the term "composable infrastructure" for a year now, but still ask why they'd want it in their data center. For some, that meaning is now clearer: On-premises hardware can help deliver infrastructure as code as part of their DevOps efforts.
Enterprises with a traditional, standard infrastructure that buy servers and blades for specific purposes through their end of life won't see much value from composable infrastructure, but it might be ideal for an agile shop, said Keith Templin, a senior systems engineer at a U.S.-based healthcare services company. He evaluated such options, in particular Synergy, at a "petting zoo" at Hewlett Packard Enterprise's (HPE) Discover event last month.
Many organizations pour significant resources into new DevOps projects "and containerizing everything," he said. "You are moving those resources around [and] consuming them in different ways for the lifecycle of that hardware."
Peyton McNullyCIO, HudsonAlpha Institute for Biotechnology
For years, IT hardware typically has been run from start to finish for one job, but composable infrastructure bounces around, carrying out different functions for various jobs. In this way, enterprises can be more like public cloud providers, giving internal customers better agility and flexibility. Traditional IT shops may need time to make the transition and get value out of composable infrastructure, Templin said.
"Everyone is trying to be like AWS [Amazon Web Services] by using the internal infrastructure and internal IT to become more like AWS in-house," he said. "Everything that DevOps stands for, they are trying to do."
HudsonAlpha Institute for Biotechnology in Huntsville, Ala., pulled development workloads out of the cloud and stood them up on bare metal on HPE Synergy.
"We treat it as another cloud and it just so happens to be bare metal on site," said CIO Peyton McNully.
This helps traditional IT "by having everyone put things on the table in a common form," using a unified API to manage software-defined infrastructure, he said.
Making composable relevant for data centers
IT pros may like the vision of composable infrastructure, but they want more practical information about how it can be deployed for their specific needs, said Dana Gardner, principal analyst at Interarbor Solutions, LLC in Gilford, N.H.
HPE "needs to bring forward logic and a path for the composable infrastructure" with more details about a disaggregated approach, he said. HPE also needs to aggressively address how hyper-converged deployments fit into a larger data center strategy with composable infrastructure.
Cisco also has pushed the composable infrastructure terminology. Its lead product, M-Series, will reach end of life later this year, but technologies developed for it will extend into other products, said James Leach, director of platform strategy for Unified Computing System (UCS) at Cisco. For example, the Cisco Virtual Interface Card technology called System Link is fabric control that came out of the M-series.
"As these capabilities move into more mainstream form factors, we see them having tremendous relevance," and Cisco's work on the M-series gave them a running start, Leach said.
Much as HPE has built Synergy around its OneView software -- a handful of customers including HudsonAlpha run Synergy in their data center -- Cisco says its strength is with the UCS control plane and the "evolutionary" addition of composable infrastructure. The company's "secret sauce" is in the automated management capability of the Cisco UCS management software, said Gina Longoria, an analyst at Moor Insights & Strategy in Austin, Texas.
Since all of Cisco's UCS products use Cisco UCS management software, the company likely will design new UCS products with disaggregated resources to take advantage of that common management infrastructure, Longoria said.
More than a name
Admittedly, the industry is still coming to grips with composable infrastructure as a definition, and many customers have not been exposed to composable infrastructure or remain unsure what it means, said Todd Brannon, director of UCS product management at Cisco. Its value will be more understood as users see ways it can help simplify their environment and eliminate complexity -- they want to see building blocks they are familiar with but that give them the ability to "dial up and dial down" versus forklift workloads, he said.
Regardless of the name, the goal is to help users develop and deploy applications quickly, he said. While they could go to the public cloud, in many cases they would prefer to do it internally, said Ric Lewis, senior vice president and general manager for converged data center infrastructure at HPE.
In addition to HPE and Cisco, Longoria expects other global vendors to begin to roll out composable infrastructure capabilities in their products during the next few years. There will be more mainstream adoption by enterprise users in the next three to five years when composability becomes a key requirement for future software-defined data centers, she said.
For now, composable infrastructure is still at the "very early stages of market development" and early adopter enterprise customers are just beginning to evaluate it in their environments.
"It is too early to declare winners and losers in this market," she said.
Robert Gates covers data centers, data center strategies, server technologies, converged and hyper-converged infrastructure and open source operating systems for SearchDataCenter. Follow him on Twitter @RBGatesTT or email him at firstname.lastname@example.org.
How disaggregation can help boost big data efficiency
Why infrastructure as code is a prerequisite for continuous delivery
HPE looks to shake up data center market with The Machine